diff --git a/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentControllerImpl.java b/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentControllerImpl.java
index 4fc15fbcd1..cc0e42d691 100644
--- a/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentControllerImpl.java
+++ b/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentControllerImpl.java
@@ -269,6 +269,7 @@ public void createOrReplaceSeldonDeployment(SeldonDeployment mlDep) {
 		    if (existing == null || !existing.getSpec().equals(mlDep.getSpec()))
 		    {
 		        logger.debug("Running updates for "+mlDep.getMetadata().getName());
+		        SeldonDeployment mlDepStatusUpdated = operator.updateStatus(mlDep);
 		        SeldonDeployment mlDep2 = operator.defaulting(mlDep);
 		        operator.validate(mlDep2);
 		        mlCache.put(mlDep2);
@@ -281,13 +282,13 @@ public void createOrReplaceSeldonDeployment(SeldonDeployment mlDep) {
 		        //removeServices(client,namespace, mlDep2, resources.services); //Proto Client not presently working for deletion
 		        ApiClient client2 = clientProvider.getClient();
 		        removeServices(client2,namespace, mlDep2, resources.services);
-		        if (!mlDep.getSpec().equals(mlDep2.getSpec()))
+		        if (!mlDep.getSpec().equals(mlDepStatusUpdated.getSpec()))
 		        {
-		            logger.debug("Pushing updated SeldonDeployment "+mlDep2.getMetadata().getName()+" back to kubectl");
-		            crdHandler.updateSeldonDeployment(mlDep2);
+		            logger.debug("Pushing updated SeldonDeployment "+mlDepStatusUpdated.getMetadata().getName()+" back to kubectl");
+		            crdHandler.updateSeldonDeployment(mlDepStatusUpdated);
 		        }
 		        else
-		            logger.debug("Not pushing an update as no change to spec for SeldonDeployment "+mlDep2.getMetadata().getName());
+		            logger.debug("Not pushing an update as no change to status for SeldonDeployment "+mlDep2.getMetadata().getName());
 		    }
 		    else
 		    {
diff --git a/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentOperator.java b/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentOperator.java
index ea12b7486a..664178928b 100644
--- a/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentOperator.java
+++ b/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentOperator.java
@@ -21,6 +21,7 @@
 
 public interface SeldonDeploymentOperator {
 	
+	public SeldonDeployment updateStatus(SeldonDeployment mlDep);
 	public SeldonDeployment defaulting(SeldonDeployment mlDep);
 	public void validate(SeldonDeployment mlDep) throws SeldonDeploymentException;
 	public DeploymentResources createResources(SeldonDeployment mlDep) throws SeldonDeploymentException;
diff --git a/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentOperatorImpl.java b/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentOperatorImpl.java
index 683807c9c9..e31ed32dd9 100644
--- a/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentOperatorImpl.java
+++ b/cluster-manager/src/main/java/io/seldon/clustermanager/k8s/SeldonDeploymentOperatorImpl.java
@@ -356,7 +356,19 @@ public String getSeldonServiceName(SeldonDeployment dep,PredictorSpec pred,Strin
 		else
 			return svcName;
 	}
-	
+
+	@Override
+	public SeldonDeployment updateStatus(SeldonDeployment mlDep) {
+		SeldonDeployment.Builder mlBuilder = SeldonDeployment.newBuilder(mlDep);
+		
+		if (!mlBuilder.hasStatus())
+		{
+			mlBuilder.getStatusBuilder().setState(Constants.STATE_CREATING);
+		}
+		
+		return mlBuilder.build();
+	}
+
 	
 	@Override
 	public SeldonDeployment defaulting(SeldonDeployment mlDep) {
@@ -378,32 +390,34 @@ public SeldonDeployment defaulting(SeldonDeployment mlDep) {
 				for(int cIdx = 0;cIdx < spec.getSpec().getContainersCount();cIdx++)
 				{
 					V1.Container c = spec.getSpec().getContainers(cIdx);
-					String containerServiceKey = getPredictorServiceNameKey(c.getName());
-					String containerServiceValue = getSeldonServiceName(mlDep, p, c.getName());
-					metaBuilder.putLabels(containerServiceKey, containerServiceValue); 
-					
-					int portNum;
-					if (servicePortMap.containsKey(c.getName()))
-						portNum = servicePortMap.get(c.getName());
-					else
+					// Only update graph and container if container is referenced in the inference graph
+					V1.Container c2;
+					if(isContainerInGraph(p.getGraph(), c))
 					{
-						portNum = currentServicePortNum;
-						servicePortMap.put(c.getName(), portNum);
-						currentServicePortNum++;
+						String containerServiceKey = getPredictorServiceNameKey(c.getName());
+						String containerServiceValue = getSeldonServiceName(mlDep, p, c.getName());
+						metaBuilder.putLabels(containerServiceKey, containerServiceValue); 
+					
+						int portNum;
+						if (servicePortMap.containsKey(c.getName()))
+							portNum = servicePortMap.get(c.getName());
+						else
+						{
+							portNum = currentServicePortNum;
+							servicePortMap.put(c.getName(), portNum);
+							currentServicePortNum++;
+						}
+						c2 = this.updateContainer(c, findPredictiveUnitForContainer(mlDep.getSpec().getPredictors(pbIdx).getGraph(),c.getName()),portNum,deploymentName,predictorName);
+						updatePredictiveUnitBuilderByName(mlBuilder.getSpecBuilder().getPredictorsBuilder(pbIdx).getGraphBuilder(),c2,containerServiceValue); 
 					}
-					V1.Container c2 = this.updateContainer(c, findPredictiveUnitForContainer(mlDep.getSpec().getPredictors(pbIdx).getGraph(),c.getName()),portNum,deploymentName,predictorName);
+					else
+						c2 = c;
 					mlBuilder.getSpecBuilder().getPredictorsBuilder(pbIdx).getComponentSpecsBuilder(ptsIdx).getSpecBuilder().addContainers(cIdx, c2);	
-					updatePredictiveUnitBuilderByName(mlBuilder.getSpecBuilder().getPredictorsBuilder(pbIdx).getGraphBuilder(),c2,containerServiceValue); 
 				}
 				mlBuilder.getSpecBuilder().getPredictorsBuilder(pbIdx).getComponentSpecsBuilder(ptsIdx).setMetadata(metaBuilder);
 			}
 		}	
 		
-		if (!mlBuilder.hasStatus())
-		{
-			mlBuilder.getStatusBuilder().setState(Constants.STATE_CREATING);
-		}
-		
 		return mlBuilder.build();
 	}
 	
@@ -504,6 +518,26 @@ private String getAmbassadorAnnotation(SeldonDeployment mlDep,String serviceName
 	    return restMapping + grpcMapping;
 	}
 	
+	/**
+	 * 
+	 * @param pu - A predictiveUnit
+	 * @param container - a container
+	 * @return True if container name can be found in graph of pu
+	 */
+	private boolean isContainerInGraph(PredictiveUnit pu,V1.Container container)
+	{
+		 if (pu.getName().equals(container.getName()))
+		 {
+	       return true;
+		 }
+		 else 
+		 {
+			 for(int i=0;i<pu.getChildrenCount();i++)
+				 if (isContainerInGraph(pu.getChildren(i),container))
+					 return true;
+		 }
+		 return false;
+	}
 	
 	private void addServicePort(PredictiveUnit pu,String serviceName,ServiceSpec.Builder svcSpecBuilder)
 	{
@@ -651,7 +685,8 @@ public DeploymentResources createResources(SeldonDeployment mlDep) throws Seldon
 					final String containerServiceKey = getPredictorServiceNameKey(c.getName());
 					final String containerServiceValue = getSeldonServiceName(mlDep, p, c.getName());
 					
-					if (!createdServices.contains(containerServiceValue))
+					// Only add a Service if container is a Seldon component in graph and we haven't already created a service for this container name
+					if (isContainerInGraph(p.getGraph(), c) && !createdServices.contains(containerServiceValue))
 					{
 						//Add service
 						Service.Builder s = Service.newBuilder()
@@ -744,5 +779,6 @@ public DeploymentResources(List<Deployment> deployments, List<Service> services)
 	}
 
 
+
 	
 }
diff --git a/docs/wrappers/nodejs.md b/docs/wrappers/nodejs.md
index 4eb5fe4c8f..e09826cf2c 100644
--- a/docs/wrappers/nodejs.md
+++ b/docs/wrappers/nodejs.md
@@ -136,13 +136,14 @@ The name of the JS file containing the model.
 
 ### API_TYPE
 
-API type to create. Can be REST only at present.
+API type to create. Can be REST or GRPC.
 
 ### SERVICE_TYPE
 
 The service type being created. Available options are:
 
 - MODEL
+- TRANSFORMER
 
 ### PERSISTENCE
 
diff --git a/examples/models/nodejs_mnist/.s2i/environment_grpc b/examples/models/nodejs_mnist/.s2i/environment_grpc
new file mode 100644
index 0000000000..cd38583149
--- /dev/null
+++ b/examples/models/nodejs_mnist/.s2i/environment_grpc
@@ -0,0 +1,4 @@
+MODEL_NAME=MnistClassifier.js
+API_TYPE=GRPC
+SERVICE_TYPE=MODEL
+PERSISTENCE=0
diff --git a/examples/models/nodejs_mnist/nodejs_mnist.ipynb b/examples/models/nodejs_mnist/nodejs_mnist.ipynb
index e2c170caea..04dc33f85a 100644
--- a/examples/models/nodejs_mnist/nodejs_mnist.ipynb
+++ b/examples/models/nodejs_mnist/nodejs_mnist.ipynb
@@ -51,7 +51,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!s2i build . seldonio/seldon-core-s2i-nodejs:0.1 node-s2i-mnist-model:0.1"
+    "!s2i build . seldonio/seldon-core-s2i-nodejs:0.2-SNAPSHOT node-s2i-mnist-model:0.1"
    ]
   },
   {
@@ -97,6 +97,65 @@
     "!docker rm nodejs_mnist_predictor --force"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Prediction using GRPC API on the docker container"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!s2i build -E ./.s2i/environment_grpc . seldonio/seldon-core-s2i-nodejs:0.2-SNAPSHOT node-s2i-mnist-model:0.2"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!docker run --name \"nodejs_mnist_predictor\" -d --rm -p 5000:5000 node-s2i-mnist-model:0.2"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!cd ../../../wrappers/testing && make build_protos"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Send some random features that conform to the contract"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!python ../../../wrappers/testing/tester.py contract.json 0.0.0.0 5000 -p -t --grpc"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!docker rm nodejs_mnist_predictor --force"
+   ]
+  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -147,7 +206,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!eval $(minikube docker-env) && s2i build . seldonio/seldon-core-s2i-nodejs:0.1 node-s2i-mnist-model:0.1"
+    "!eval $(minikube docker-env) && s2i build . seldonio/seldon-core-s2i-nodejs:0.2-SNAPSHOT node-s2i-mnist-model:0.1"
    ]
   },
   {
diff --git a/examples/models/nodejs_tensorflow/.s2i/environment_grpc b/examples/models/nodejs_tensorflow/.s2i/environment_grpc
new file mode 100644
index 0000000000..b50695cae7
--- /dev/null
+++ b/examples/models/nodejs_tensorflow/.s2i/environment_grpc
@@ -0,0 +1,4 @@
+MODEL_NAME=MyModel.js
+API_TYPE=GRPC
+SERVICE_TYPE=MODEL
+PERSISTENCE=0
diff --git a/examples/models/nodejs_tensorflow/nodejs_tensorflow.ipynb b/examples/models/nodejs_tensorflow/nodejs_tensorflow.ipynb
index bfbe083a45..f2135949d3 100644
--- a/examples/models/nodejs_tensorflow/nodejs_tensorflow.ipynb
+++ b/examples/models/nodejs_tensorflow/nodejs_tensorflow.ipynb
@@ -50,7 +50,7 @@
 			"metadata": {},
 			"outputs": [],
 			"source": [
-				"!s2i build . seldonio/seldon-core-s2i-nodejs:0.1 node-s2i-model-image:0.1"
+				"!s2i build . seldonio/seldon-core-s2i-nodejs:0.2-SNAPSHOT node-s2i-model-image:0.1"
 			]
 		},
 		{
@@ -96,6 +96,72 @@
 				"!docker rm nodejs_tensorflow_predictor --force"
 			]
 		},
+		{
+			"cell_type": "markdown",
+			"metadata": {},
+			"source": [
+				"## Prediction using GRPC API on the docker container"
+			]
+		},
+		{
+			"cell_type": "code",
+			"execution_count": null,
+			"metadata": {},
+			"outputs": [],
+			"source": [
+				"!s2i build -E ./.s2i/environment_grpc . seldonio/seldon-core-s2i-nodejs:0.2-SNAPSHOT node-s2i-model-image:0.2"
+			]
+		},
+		{
+			"cell_type": "code",
+			"execution_count": null,
+			"metadata": {},
+			"outputs": [],
+			"source": [
+				"!docker run --name \"nodejs_tensorflow_predictor\" -d --rm -p 5000:5000 node-s2i-model-image:0.2"
+			]
+		},
+		{
+			"cell_type": "code",
+			"execution_count": null,
+			"metadata": {},
+			"outputs": [],
+			"source": [
+				"!cd ../../../wrappers/testing && make build_protos"
+			]
+		},
+		{
+			"cell_type": "markdown",
+			"metadata": {},
+			"source": [
+				"Send some random features that conform to the contract"
+			]
+		},
+		{
+			"cell_type": "code",
+			"execution_count": null,
+			"metadata": {},
+			"outputs": [],
+			"source": [
+				"!python ../../../wrappers/testing/tester.py contract.json 0.0.0.0 5000 -p -t --grpc"
+			]
+		},
+		{
+			"cell_type": "code",
+			"execution_count": null,
+			"metadata": {},
+			"outputs": [],
+			"source": [
+				"!docker rm nodejs_tensorflow_predictor --force"
+			]
+		},
+		{
+			"cell_type": "markdown",
+			"metadata": {},
+			"source": [
+				"## Prediction using Minikube"
+			]
+		},
 		{
 			"cell_type": "code",
 			"execution_count": null,
@@ -139,7 +205,7 @@
 			"metadata": {},
 			"outputs": [],
 			"source": [
-				"!eval $(minikube docker-env) && s2i build . seldonio/seldon-core-s2i-nodejs:0.1 node-s2i-model-image:0.1"
+				"!eval $(minikube docker-env) && s2i build . seldonio/seldon-core-s2i-nodejs:0.2-SNAPSHOT node-s2i-model-image:0.1"
 			]
 		},
 		{
diff --git a/examples/models/nvidia-mnist/.s2i/environment b/examples/models/nvidia-mnist/.s2i/environment
new file mode 100644
index 0000000000..752bf2fd05
--- /dev/null
+++ b/examples/models/nvidia-mnist/.s2i/environment
@@ -0,0 +1,4 @@
+MODEL_NAME=MnistTransformer
+API_TYPE=REST
+SERVICE_TYPE=TRANSFORMER
+PERSISTENCE=0
diff --git a/examples/models/nvidia-mnist/Makefile b/examples/models/nvidia-mnist/Makefile
new file mode 100644
index 0000000000..25e5552595
--- /dev/null
+++ b/examples/models/nvidia-mnist/Makefile
@@ -0,0 +1,14 @@
+TRANSFORMER_IMAGE=seldonio/mnist-caffe2-transformer:0.1
+
+clean:
+	rm -f rm -f tensorrt_mnist/1/model.plan
+	rm -rf MNIST_data
+	rm -f mnist.json
+	rm -f tmp.json
+
+build_transformer:
+	s2i build . seldonio/seldon-core-s2i-python3:0.2 ${TRANSFORMER_IMAGE}
+
+push_transformer:
+	docker push ${TRANSFORMER_IMAGE}
+
diff --git a/examples/models/nvidia-mnist/MnistTransformer.py b/examples/models/nvidia-mnist/MnistTransformer.py
new file mode 100644
index 0000000000..84b097a077
--- /dev/null
+++ b/examples/models/nvidia-mnist/MnistTransformer.py
@@ -0,0 +1,27 @@
+import numpy as np
+
+MEANS=np.array([255.0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,254,254,254,253,252,252,251,251,252,252,253,254,254,255,255,255,255,255,255,255,255,255,255,255,255,255,254,254,253,251,249,248,245,243,242,242,243,246,248,251,253,254,255,255,255,255,255,255,255,255,255,255,255,254,253,250,247,242,235,228,220,213,210,211,216,224,232,240,246,251,253,254,255,255,255,255,255,255,255,255,254,251,248,242,234,223,211,196,181,170,164,166,175,189,205,221,233,243,248,252,254,255,255,255,255,255,255,254,252,248,241,231,217,202,184,166,149,136,131,134,143,159,180,201,220,234,243,249,253,255,255,255,255,255,254,253,249,243,233,219,201,181,161,143,130,122,120,122,129,141,161,185,208,227,240,248,252,254,255,255,255,255,254,251,246,238,226,208,187,164,146,135,131,132,133,132,133,139,154,178,202,223,239,248,252,255,255,255,255,254,253,251,245,236,221,200,177,156,144,144,150,156,156,151,144,144,156,178,202,224,240,249,253,255,255,255,255,254,253,251,245,235,218,195,172,155,152,161,172,176,170,161,150,149,161,183,207,227,242,250,254,255,255,255,255,255,254,251,246,234,215,191,168,156,160,173,182,179,169,157,147,149,166,190,213,230,243,251,254,255,255,255,255,255,254,252,246,233,212,186,165,157,164,175,176,165,153,142,137,147,170,196,217,231,242,251,255,255,255,255,255,255,254,252,245,230,207,182,163,158,164,168,158,143,131,125,128,146,174,200,218,231,241,250,254,255,255,255,255,255,255,252,243,227,205,181,164,159,161,157,139,124,115,118,127,148,176,199,216,230,240,249,254,255,255,255,255,255,254,251,241,224,204,184,169,163,160,150,132,119,116,123,133,153,177,197,214,228,240,249,254,255,255,255,255,255,254,251,239,222,205,189,177,171,166,154,139,129,128,134,144,159,177,195,213,228,241,249,254,255,255,255,255,255,254,249,237,222,207,195,186,180,175,166,153,143,140,142,150,162,178,195,214,230,242,250,254,255,255,255,255,255,253,247,235,220,207,197,189,183,179,172,160,148,142,143,150,161,178,198,217,233,244,250,254,255,255,255,255,255,253,246,233,218,204,192,184,177,172,165,153,142,137,139,148,163,183,204,222,236,246,251,254,255,255,255,255,255,253,247,234,218,201,186,174,165,157,148,137,130,129,137,151,171,194,214,230,242,248,252,254,255,255,255,255,255,253,249,238,222,203,184,168,154,143,132,124,123,130,145,165,188,209,227,239,247,251,253,255,255,255,255,255,255,254,251,244,232,214,194,174,156,142,132,130,134,148,167,189,210,226,238,246,250,253,254,255,255,255,255,255,255,255,253,250,243,231,215,196,178,163,155,156,164,179,197,215,230,240,247,251,253,254,255,255,255,255,255,255,255,255,254,253,251,246,238,228,217,208,203,204,210,218,228,236,243,248,251,253,254,255,255,255,255,255,255,255,255,255,255,255,254,252,249,245,241,238,237,237,239,242,245,247,250,252,253,254,255,255,255,255,255,255,255,255,255,255,255,255,254,254,253,252,250,249,248,249,249,250,252,253,253,254,254,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,254,254,254,254,255,255,255,255,255,255,255,255,255,255,255,255])
+
+
+class MnistTransformer(object):
+
+    def __init__(self):
+        print("init");
+        
+    def preProcessMNIST(self,X):
+        '''
+        Convert values assumed to be in 0-1 range to a value in 0-255.
+        The remove the training mean needed by the Caffe2 model.
+        Finally reshape the output to that expected by the model
+        '''
+        X = X * 255
+        X = 255 - X
+        X = (X.reshape(784) - MEANS).reshape(28,28,1)
+        X = np.transpose(X, (2, 0, 1))
+        return X
+
+    def transform_input(self,X,names):
+        return self.preProcessMNIST(X)
+
+    def transform_output(self,X,names):
+        return X.reshape(1,10)
diff --git a/examples/models/nvidia-mnist/fetch-model.sh b/examples/models/nvidia-mnist/fetch-model.sh
new file mode 100755
index 0000000000..6b298153f6
--- /dev/null
+++ b/examples/models/nvidia-mnist/fetch-model.sh
@@ -0,0 +1 @@
+wget -O mnist_tensorrt_model/1/model.plan http://seldon-public.s3.amazonaws.com/nvidia-mnist-model/model.plan
diff --git a/examples/models/nvidia-mnist/nvidia-mnist/Chart.yaml b/examples/models/nvidia-mnist/nvidia-mnist/Chart.yaml
new file mode 100644
index 0000000000..cc00617700
--- /dev/null
+++ b/examples/models/nvidia-mnist/nvidia-mnist/Chart.yaml
@@ -0,0 +1,6 @@
+apiVersion: v1
+description: Seldon MNIST Nvidia Inference Server Example
+name: nvidia-mnist
+sources:
+- https://github.com/SeldonIO/seldon-core
+version: 0.1
diff --git a/examples/models/nvidia-mnist/nvidia-mnist/templates/mnist_nvidia_deployment.json b/examples/models/nvidia-mnist/nvidia-mnist/templates/mnist_nvidia_deployment.json
new file mode 100644
index 0000000000..97ce30c6fe
--- /dev/null
+++ b/examples/models/nvidia-mnist/nvidia-mnist/templates/mnist_nvidia_deployment.json
@@ -0,0 +1,135 @@
+{
+    "apiVersion": "machinelearning.seldon.io/v1alpha2",
+    "kind": "SeldonDeployment",
+    "metadata": {
+        "labels": {
+            "app": "seldon"
+        },
+        "name": "nvidia-mnist",
+	"namespace": "{{ .Release.Namespace }}"
+    },
+    "spec": {
+        "name": "caffe2-mnist",
+        "predictors": [
+            {
+                "componentSpecs": [{
+                    "spec": {
+                        "containers": [
+                            {
+                                "image": "seldonio/mnist-caffe2-transformer:0.1",
+                                "name": "mnist-transformer"
+                            },
+                            {
+                                "image": "seldonio/nvidia-inference-server-proxy:0.1",
+                                "name": "nvidia-proxy"
+                            },
+			    {
+				"args": [
+				    "--model-store={{ .Values.nvidia.model_store }}"
+				],
+				"command": [
+				    "inference_server"
+				],
+				"image": "nvcr.io/nvidia/inferenceserver:18.08.1-py2",
+				"livenessProbe": {
+				    "failureThreshold": 3,
+				    "handler":{
+					"httpGet": {
+					    "path": "/api/health/live",
+					    "port": {{ .Values.nvidia.port }},
+					    "scheme": "HTTP"
+					}
+				    },
+				    "initialDelaySeconds": 5,
+				    "periodSeconds": 5,
+				    "successThreshold": 1,
+				    "timeoutSeconds": 1
+				},
+				"name": "inference-server",
+				"ports": [
+				    {
+					"containerPort": {{ .Values.nvidia.port }},
+					"protocol": "TCP"
+				    },
+				    {
+					"containerPort": 8001,
+					"protocol": "TCP"
+				    },
+				    {
+					"containerPort": 8002,
+					"protocol": "TCP"
+				    }
+				],
+				"readinessProbe": {
+				    "failureThreshold": 3,
+				    "handler":{
+					"httpGet": {
+					    "path": "/api/health/ready",
+					    "port": {{ .Values.nvidia.port }},
+					    "scheme": "HTTP"
+					}
+				    },
+				    "initialDelaySeconds": 5,
+				    "periodSeconds": 5,
+				    "successThreshold": 1,
+				    "timeoutSeconds": 1
+				},
+				"resources": {
+				    "limits": {
+					"nvidia.com/gpu": "1"
+				    },
+				    "requests": {
+					"cpu": "100m",
+					"nvidia.com/gpu": "1"
+				    }
+				},
+				"securityContext": {
+				    "runAsUser": 1000
+				}
+			    }
+			],
+			"terminationGracePeriodSeconds": 1,
+			"imagePullSecrets": [
+			    {
+				"name": "ngc"
+			    }
+			]
+		    }
+		}],
+                "graph": {
+                    "name": "mnist-transformer",
+                    "endpoint": { "type" : "REST" },
+                    "type": "TRANSFORMER",
+		    "children": [
+			{
+			    "name": "nvidia-proxy",
+			    "endpoint": { "type" : "REST" },
+			    "type": "MODEL",
+			    "children": [],
+			    "parameters":
+			    [
+				{
+				    "name":"url",
+				    "type":"STRING",
+				    "value":"127.0.0.1:{{ .Values.nvidia.port }}"
+				},
+				{
+				    "name":"model_name",
+				    "type":"STRING",
+				    "value":"tensorrt_mnist"
+				},
+				{
+				    "name":"protocol",
+				    "type":"STRING",
+				    "value":"HTTP"
+				}
+			    ]
+			}
+		    ]
+                },
+                "name": "mnist-nvidia",
+                "replicas": 1
+            }
+        ]
+    }
+}
diff --git a/examples/models/nvidia-mnist/nvidia-mnist/values.yaml b/examples/models/nvidia-mnist/nvidia-mnist/values.yaml
new file mode 100644
index 0000000000..adaff98e15
--- /dev/null
+++ b/examples/models/nvidia-mnist/nvidia-mnist/values.yaml
@@ -0,0 +1,4 @@
+nvidia:
+  model_store: gs://seldon-inference-server-model-store
+  port: 8000
+  
diff --git a/examples/models/nvidia-mnist/nvidia_mnist.ipynb b/examples/models/nvidia-mnist/nvidia_mnist.ipynb
new file mode 100644
index 0000000000..c20b80d5f8
--- /dev/null
+++ b/examples/models/nvidia-mnist/nvidia_mnist.ipynb
@@ -0,0 +1,1361 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Seldon and NVIDIA Inference Server MNIST Example\n",
+    "\n",
+    "This example shows how you can combine Seldon with the NVIDIA Inference Server. We will use a Seldon TensorRT proxy model image that will forward Seldon internal microservice prediction calls out to an external TensorRT Inference Server.\n",
+    "\n",
+    "The example will use the MNIST digit classification task with a pre-trained CAFFE2 model.\n",
+    "\n",
+    "A Seldon transformer will transform the inputs before sending to the Proxy which will forward the request to the Nvidia Inference Server.\n",
+    "\n",
+    "This example will:\n",
+    "\n",
+    " * Show the packaging of the components using S2I and a step by step local testing of these via Docker\n",
+    " * Show running the example in Seldon Core on GCP with an embedded Nvidia Inference Server\n",
+    " "
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Setup"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/home/clive/anaconda3/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
+      "  from ._conv import register_converters as _register_converters\n"
+     ]
+    }
+   ],
+   "source": [
+    "%matplotlib inline\n",
+    "import requests\n",
+    "from random import randint,random\n",
+    "import json\n",
+    "from matplotlib import pyplot as plt\n",
+    "import numpy as np\n",
+    "from tensorflow.examples.tutorials.mnist import input_data\n",
+    "import sys\n",
+    "sys.path.append(\"../../../notebooks\")\n",
+    "from visualizer import get_graph"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def gen_image(arr):\n",
+    "    two_d = (np.reshape(arr, (28, 28)) * 255).astype(np.uint8)\n",
+    "    plt.imshow(two_d,cmap=plt.cm.gray_r, interpolation='nearest')\n",
+    "    return plt\n",
+    "\n",
+    "def download_mnist():\n",
+    "    return input_data.read_data_sets(\"MNIST_data/\", one_hot = True)\n",
+    "\n",
+    "def rest_predict_request(endpoint,data):\n",
+    "    request = {\"data\":{\"ndarray\":data.tolist()}}\n",
+    "    response = requests.post(\n",
+    "                \"http://\"+endpoint+\"/predict\",\n",
+    "                data={\"json\":json.dumps(request),\"isDefault\":True})\n",
+    "    return response.json()   \n",
+    "\n",
+    "def rest_transform_input_request(endpoint,data):\n",
+    "    request = {\"data\":{\"ndarray\":data.tolist()}}\n",
+    "    response = requests.post(\n",
+    "                \"http://\"+endpoint+\"/transform-input\",\n",
+    "                data={\"json\":json.dumps(request),\"isDefault\":True})\n",
+    "    return response.json()   \n",
+    "\n",
+    "def rest_transform_output_request(endpoint,data):\n",
+    "    request = {\"data\":{\"ndarray\":data.tolist()}}\n",
+    "    response = requests.post(\n",
+    "                \"http://\"+endpoint+\"/transform-output\",\n",
+    "                data={\"json\":json.dumps(request),\"isDefault\":True})\n",
+    "    return response.json()   \n",
+    "\n",
+    "def rest_request_ambassador(deploymentName,endpoint=\"localhost:8003\",arr=None):\n",
+    "    payload = {\"data\":{\"names\":[\"a\",\"b\"],\"tensor\":{\"shape\":[1,784],\"values\":arr.tolist()}}}\n",
+    "    response = requests.post(\n",
+    "                \"http://\"+endpoint+\"/seldon/\"+deploymentName+\"/api/v0.1/predictions\",\n",
+    "                json=payload)\n",
+    "    print(response.status_code)\n",
+    "    print(response.text)\n",
+    "\n",
+    "\n",
+    "def gen_mnist_data(mnist):\n",
+    "    batch_xs, batch_ys = mnist.train.next_batch(1)\n",
+    "    chosen=0\n",
+    "    gen_image(batch_xs[chosen]).show()\n",
+    "    data = batch_xs[chosen].reshape((1,784))\n",
+    "    return data\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "WARNING:tensorflow:From <ipython-input-2-8029beac92e7>:7: read_data_sets (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use alternatives such as official/mnist/dataset.py from tensorflow/models.\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:260: maybe_download (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please write your own downloading logic.\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/base.py:252: _internal_retry.<locals>.wrap.<locals>.wrapped_fn (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use urllib or similar directly.\n",
+      "Successfully downloaded train-images-idx3-ubyte.gz 9912422 bytes.\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:262: extract_images (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use tf.data to implement this functionality.\n",
+      "Extracting MNIST_data/train-images-idx3-ubyte.gz\n",
+      "Successfully downloaded train-labels-idx1-ubyte.gz 28881 bytes.\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:267: extract_labels (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use tf.data to implement this functionality.\n",
+      "Extracting MNIST_data/train-labels-idx1-ubyte.gz\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:110: dense_to_one_hot (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use tf.one_hot on tensors.\n",
+      "Successfully downloaded t10k-images-idx3-ubyte.gz 1648877 bytes.\n",
+      "Extracting MNIST_data/t10k-images-idx3-ubyte.gz\n",
+      "Successfully downloaded t10k-labels-idx1-ubyte.gz 4542 bytes.\n",
+      "Extracting MNIST_data/t10k-labels-idx1-ubyte.gz\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:290: DataSet.__init__ (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use alternatives such as official/mnist/dataset.py from tensorflow/models.\n"
+     ]
+    }
+   ],
+   "source": [
+    "mnist = download_mnist()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Create an Nvidia Model Repository"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Fetch pretrained MNIST model ready for serving and place in model repository"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "--2018-09-20 14:18:52--  http://seldon-public.s3.amazonaws.com/nvidia-mnist-model/model.plan\n",
+      "Resolving seldon-public.s3.amazonaws.com (seldon-public.s3.amazonaws.com)... 52.218.48.249\n",
+      "Connecting to seldon-public.s3.amazonaws.com (seldon-public.s3.amazonaws.com)|52.218.48.249|:80... connected.\n",
+      "HTTP request sent, awaiting response... 200 OK\n",
+      "Length: 1731864 (1.7M) [binary/octet-stream]\n",
+      "Saving to: ‘tensorrt_mnist/1/model.plan’\n",
+      "\n",
+      "tensorrt_mnist/1/mo 100%[===================>]   1.65M   148KB/s    in 12s     \n",
+      "\n",
+      "2018-09-20 14:19:05 (136 KB/s) - ‘tensorrt_mnist/1/model.plan’ saved [1731864/1731864]\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "!wget -O tensorrt_mnist/1/model.plan http://seldon-public.s3.amazonaws.com/nvidia-mnist-model/model.plan"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "To run your Mvidia Inference Server you will need to upload to a google bucket the model repository in mnsit_tensorrt_model. Follow the steps below:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "env: MODEL_REPOSITORY_BUCKET=gs://seldon-inference-server-model-store\n"
+     ]
+    }
+   ],
+   "source": [
+    "# CHANGE THIS TO YOUR OWN CHOSEN GOOGLE BUCKET NAME\n",
+    "%env MODEL_REPOSITORY_BUCKET=gs://seldon-inference-server-model-store"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Creating gs://seldon-inference-server-model-store/...\n",
+      "ServiceException: 409 Bucket seldon-inference-server-model-store already exists.\n"
+     ]
+    }
+   ],
+   "source": [
+    "!gsutil mb ${MODEL_REPOSITORY_BUCKET}"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Copying file://tensorrt_mnist/config.pbtxt [Content-Type=application/octet-stream]...\n",
+      "Copying file://tensorrt_mnist/config.pbtxt~ [Content-Type=application/octet-stream]...\n",
+      "Copying file://tensorrt_mnist/mnist_labels.txt [Content-Type=text/plain]...     \n",
+      "Copying file://tensorrt_mnist/1/model.plan [Content-Type=application/octet-stream]...\n",
+      "\\ [4 files][  1.7 MiB/  1.7 MiB]                                                \n",
+      "Operation completed over 4 objects/1.7 MiB.                                      \n"
+     ]
+    }
+   ],
+   "source": [
+    "!gsutil cp -r tensorrt_mnist ${MODEL_REPOSITORY_BUCKET}"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Test Locally From Docker with Standalone NVIDIA Inference Server"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Create your Nvidia Inference Server\n",
+    "\n",
+    "For example:\n",
+    "\n",
+    " * Follow the steps in the [Kubeflow guide](https://github.com/kubeflow/kubeflow/tree/master/kubeflow/nvidia-inference-server) to create your Nvidia Inference Server \n",
+    "   * You will need to use the Google Bucket location"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Package and run a Transformer and Nvidia Proxy"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "We will use a Seldon Transformer to remove the training-set mean values from the input features and rehsape the output as the prediction comes back."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\u001b[34mimport\u001b[39;49;00m \u001b[04m\u001b[36mnumpy\u001b[39;49;00m \u001b[34mas\u001b[39;49;00m \u001b[04m\u001b[36mnp\u001b[39;49;00m\r\n",
+      "\r\n",
+      "MEANS=np.array([\u001b[34m255.0\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m248\u001b[39;49;00m,\u001b[34m245\u001b[39;49;00m,\u001b[34m243\u001b[39;49;00m,\u001b[34m242\u001b[39;49;00m,\u001b[34m242\u001b[39;49;00m,\u001b[34m243\u001b[39;49;00m,\u001b[34m246\u001b[39;49;00m,\u001b[34m248\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m250\u001b[39;49;00m,\u001b[34m247\u001b[39;49;00m,\u001b[34m242\u001b[39;49;00m,\u001b[34m235\u001b[39;49;00m,\u001b[34m228\u001b[39;49;00m,\u001b[34m220\u001b[39;49;00m,\u001b[34m213\u001b[39;49;00m,\u001b[34m210\u001b[39;49;00m,\u001b[34m211\u001b[39;49;00m,\u001b[34m216\u001b[39;49;00m,\u001b[34m224\u001b[39;49;00m,\u001b[34m232\u001b[39;49;00m,\u001b[34m240\u001b[39;49;00m,\u001b[34m246\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m248\u001b[39;49;00m,\u001b[34m242\u001b[39;49;00m,\u001b[34m234\u001b[39;49;00m,\u001b[34m223\u001b[39;49;00m,\u001b[34m211\u001b[39;49;00m,\u001b[34m196\u001b[39;49;00m,\u001b[34m181\u001b[39;49;00m,\u001b[34m170\u001b[39;49;00m,\u001b[34m164\u001b[39;49;00m,\u001b[34m166\u001b[39;49;00m,\u001b[34m175\u001b[39;49;00m,\u001b[34m189\u001b[39;49;00m,\u001b[34m205\u001b[39;49;00m,\u001b[34m221\u001b[39;49;00m,\u001b[34m233\u001b[39;49;00m,\u001b[34m243\u001b[39;49;00m,\u001b[34m248\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m248\u001b[39;49;00m,\u001b[34m241\u001b[39;49;00m,\u001b[34m231\u001b[39;49;00m,\u001b[34m217\u001b[39;49;00m,\u001b[34m202\u001b[39;49;00m,\u001b[34m184\u001b[39;49;00m,\u001b[34m166\u001b[39;49;00m,\u001b[34m149\u001b[39;49;00m,\u001b[34m136\u001b[39;49;00m,\u001b[34m131\u001b[39;49;00m,\u001b[34m134\u001b[39;49;00m,\u001b[34m143\u001b[39;49;00m,\u001b[34m159\u001b[39;49;00m,\u001b[34m180\u001b[39;49;00m,\u001b[34m201\u001b[39;49;00m,\u001b[34m220\u001b[39;49;00m,\u001b[34m234\u001b[39;49;00m,\u001b[34m243\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m243\u001b[39;49;00m,\u001b[34m233\u001b[39;49;00m,\u001b[34m219\u001b[39;49;00m,\u001b[34m201\u001b[39;49;00m,\u001b[34m181\u001b[39;49;00m,\u001b[34m161\u001b[39;49;00m,\u001b[34m143\u001b[39;49;00m,\u001b[34m130\u001b[39;49;00m,\u001b[34m122\u001b[39;49;00m,\u001b[34m120\u001b[39;49;00m,\u001b[34m122\u001b[39;49;00m,\u001b[34m129\u001b[39;49;00m,\u001b[34m141\u001b[39;49;00m,\u001b[34m161\u001b[39;49;00m,\u001b[34m185\u001b[39;49;00m,\u001b[34m208\u001b[39;49;00m,\u001b[34m227\u001b[39;49;00m,\u001b[34m240\u001b[39;49;00m,\u001b[34m248\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m246\u001b[39;49;00m,\u001b[34m238\u001b[39;49;00m,\u001b[34m226\u001b[39;49;00m,\u001b[34m208\u001b[39;49;00m,\u001b[34m187\u001b[39;49;00m,\u001b[34m164\u001b[39;49;00m,\u001b[34m146\u001b[39;49;00m,\u001b[34m135\u001b[39;49;00m,\u001b[34m131\u001b[39;49;00m,\u001b[34m132\u001b[39;49;00m,\u001b[34m133\u001b[39;49;00m,\u001b[34m132\u001b[39;49;00m,\u001b[34m133\u001b[39;49;00m,\u001b[34m139\u001b[39;49;00m,\u001b[34m154\u001b[39;49;00m,\u001b[34m178\u001b[39;49;00m,\u001b[34m202\u001b[39;49;00m,\u001b[34m223\u001b[39;49;00m,\u001b[34m239\u001b[39;49;00m,\u001b[34m248\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m245\u001b[39;49;00m,\u001b[34m236\u001b[39;49;00m,\u001b[34m221\u001b[39;49;00m,\u001b[34m200\u001b[39;49;00m,\u001b[34m177\u001b[39;49;00m,\u001b[34m156\u001b[39;49;00m,\u001b[34m144\u001b[39;49;00m,\u001b[34m144\u001b[39;49;00m,\u001b[34m150\u001b[39;49;00m,\u001b[34m156\u001b[39;49;00m,\u001b[34m156\u001b[39;49;00m,\u001b[34m151\u001b[39;49;00m,\u001b[34m144\u001b[39;49;00m,\u001b[34m144\u001b[39;49;00m,\u001b[34m156\u001b[39;49;00m,\u001b[34m178\u001b[39;49;00m,\u001b[34m202\u001b[39;49;00m,\u001b[34m224\u001b[39;49;00m,\u001b[34m240\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m245\u001b[39;49;00m,\u001b[34m235\u001b[39;49;00m,\u001b[34m218\u001b[39;49;00m,\u001b[34m195\u001b[39;49;00m,\u001b[34m172\u001b[39;49;00m,\u001b[34m155\u001b[39;49;00m,\u001b[34m152\u001b[39;49;00m,\u001b[34m161\u001b[39;49;00m,\u001b[34m172\u001b[39;49;00m,\u001b[34m176\u001b[39;49;00m,\u001b[34m170\u001b[39;49;00m,\u001b[34m161\u001b[39;49;00m,\u001b[34m150\u001b[39;49;00m,\u001b[34m149\u001b[39;49;00m,\u001b[34m161\u001b[39;49;00m,\u001b[34m183\u001b[39;49;00m,\u001b[34m207\u001b[39;49;00m,\u001b[34m227\u001b[39;49;00m,\u001b[34m242\u001b[39;49;00m,\u001b[34m250\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m246\u001b[39;49;00m,\u001b[34m234\u001b[39;49;00m,\u001b[34m215\u001b[39;49;00m,\u001b[34m191\u001b[39;49;00m,\u001b[34m168\u001b[39;49;00m,\u001b[34m156\u001b[39;49;00m,\u001b[34m160\u001b[39;49;00m,\u001b[34m173\u001b[39;49;00m,\u001b[34m182\u001b[39;49;00m,\u001b[34m179\u001b[39;49;00m,\u001b[34m169\u001b[39;49;00m,\u001b[34m157\u001b[39;49;00m,\u001b[34m147\u001b[39;49;00m,\u001b[34m149\u001b[39;49;00m,\u001b[34m166\u001b[39;49;00m,\u001b[34m190\u001b[39;49;00m,\u001b[34m213\u001b[39;49;00m,\u001b[34m230\u001b[39;49;00m,\u001b[34m243\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m246\u001b[39;49;00m,\u001b[34m233\u001b[39;49;00m,\u001b[34m212\u001b[39;49;00m,\u001b[34m186\u001b[39;49;00m,\u001b[34m165\u001b[39;49;00m,\u001b[34m157\u001b[39;49;00m,\u001b[34m164\u001b[39;49;00m,\u001b[34m175\u001b[39;49;00m,\u001b[34m176\u001b[39;49;00m,\u001b[34m165\u001b[39;49;00m,\u001b[34m153\u001b[39;49;00m,\u001b[34m142\u001b[39;49;00m,\u001b[34m137\u001b[39;49;00m,\u001b[34m147\u001b[39;49;00m,\u001b[34m170\u001b[39;49;00m,\u001b[34m196\u001b[39;49;00m,\u001b[34m217\u001b[39;49;00m,\u001b[34m231\u001b[39;49;00m,\u001b[34m242\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m245\u001b[39;49;00m,\u001b[34m230\u001b[39;49;00m,\u001b[34m207\u001b[39;49;00m,\u001b[34m182\u001b[39;49;00m,\u001b[34m163\u001b[39;49;00m,\u001b[34m158\u001b[39;49;00m,\u001b[34m164\u001b[39;49;00m,\u001b[34m168\u001b[39;49;00m,\u001b[34m158\u001b[39;49;00m,\u001b[34m143\u001b[39;49;00m,\u001b[34m131\u001b[39;49;00m,\u001b[34m125\u001b[39;49;00m,\u001b[34m128\u001b[39;49;00m,\u001b[34m146\u001b[39;49;00m,\u001b[34m174\u001b[39;49;00m,\u001b[34m200\u001b[39;49;00m,\u001b[34m218\u001b[39;49;00m,\u001b[34m231\u001b[39;49;00m,\u001b[34m241\u001b[39;49;00m,\u001b[34m250\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m243\u001b[39;49;00m,\u001b[34m227\u001b[39;49;00m,\u001b[34m205\u001b[39;49;00m,\u001b[34m181\u001b[39;49;00m,\u001b[34m164\u001b[39;49;00m,\u001b[34m159\u001b[39;49;00m,\u001b[34m161\u001b[39;49;00m,\u001b[34m157\u001b[39;49;00m,\u001b[34m139\u001b[39;49;00m,\u001b[34m124\u001b[39;49;00m,\u001b[34m115\u001b[39;49;00m,\u001b[34m118\u001b[39;49;00m,\u001b[34m127\u001b[39;49;00m,\u001b[34m148\u001b[39;49;00m,\u001b[34m176\u001b[39;49;00m,\u001b[34m199\u001b[39;49;00m,\u001b[34m216\u001b[39;49;00m,\u001b[34m230\u001b[39;49;00m,\u001b[34m240\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m241\u001b[39;49;00m,\u001b[34m224\u001b[39;49;00m,\u001b[34m204\u001b[39;49;00m,\u001b[34m184\u001b[39;49;00m,\u001b[34m169\u001b[39;49;00m,\u001b[34m163\u001b[39;49;00m,\u001b[34m160\u001b[39;49;00m,\u001b[34m150\u001b[39;49;00m,\u001b[34m132\u001b[39;49;00m,\u001b[34m119\u001b[39;49;00m,\u001b[34m116\u001b[39;49;00m,\u001b[34m123\u001b[39;49;00m,\u001b[34m133\u001b[39;49;00m,\u001b[34m153\u001b[39;49;00m,\u001b[34m177\u001b[39;49;00m,\u001b[34m197\u001b[39;49;00m,\u001b[34m214\u001b[39;49;00m,\u001b[34m228\u001b[39;49;00m,\u001b[34m240\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m239\u001b[39;49;00m,\u001b[34m222\u001b[39;49;00m,\u001b[34m205\u001b[39;49;00m,\u001b[34m189\u001b[39;49;00m,\u001b[34m177\u001b[39;49;00m,\u001b[34m171\u001b[39;49;00m,\u001b[34m166\u001b[39;49;00m,\u001b[34m154\u001b[39;49;00m,\u001b[34m139\u001b[39;49;00m,\u001b[34m129\u001b[39;49;00m,\u001b[34m128\u001b[39;49;00m,\u001b[34m134\u001b[39;49;00m,\u001b[34m144\u001b[39;49;00m,\u001b[34m159\u001b[39;49;00m,\u001b[34m177\u001b[39;49;00m,\u001b[34m195\u001b[39;49;00m,\u001b[34m213\u001b[39;49;00m,\u001b[34m228\u001b[39;49;00m,\u001b[34m241\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m237\u001b[39;49;00m,\u001b[34m222\u001b[39;49;00m,\u001b[34m207\u001b[39;49;00m,\u001b[34m195\u001b[39;49;00m,\u001b[34m186\u001b[39;49;00m,\u001b[34m180\u001b[39;49;00m,\u001b[34m175\u001b[39;49;00m,\u001b[34m166\u001b[39;49;00m,\u001b[34m153\u001b[39;49;00m,\u001b[34m143\u001b[39;49;00m,\u001b[34m140\u001b[39;49;00m,\u001b[34m142\u001b[39;49;00m,\u001b[34m150\u001b[39;49;00m,\u001b[34m162\u001b[39;49;00m,\u001b[34m178\u001b[39;49;00m,\u001b[34m195\u001b[39;49;00m,\u001b[34m214\u001b[39;49;00m,\u001b[34m230\u001b[39;49;00m,\u001b[34m242\u001b[39;49;00m,\u001b[34m250\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m247\u001b[39;49;00m,\u001b[34m235\u001b[39;49;00m,\u001b[34m220\u001b[39;49;00m,\u001b[34m207\u001b[39;49;00m,\u001b[34m197\u001b[39;49;00m,\u001b[34m189\u001b[39;49;00m,\u001b[34m183\u001b[39;49;00m,\u001b[34m179\u001b[39;49;00m,\u001b[34m172\u001b[39;49;00m,\u001b[34m160\u001b[39;49;00m,\u001b[34m148\u001b[39;49;00m,\u001b[34m142\u001b[39;49;00m,\u001b[34m143\u001b[39;49;00m,\u001b[34m150\u001b[39;49;00m,\u001b[34m161\u001b[39;49;00m,\u001b[34m178\u001b[39;49;00m,\u001b[34m198\u001b[39;49;00m,\u001b[34m217\u001b[39;49;00m,\u001b[34m233\u001b[39;49;00m,\u001b[34m244\u001b[39;49;00m,\u001b[34m250\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m246\u001b[39;49;00m,\u001b[34m233\u001b[39;49;00m,\u001b[34m218\u001b[39;49;00m,\u001b[34m204\u001b[39;49;00m,\u001b[34m192\u001b[39;49;00m,\u001b[34m184\u001b[39;49;00m,\u001b[34m177\u001b[39;49;00m,\u001b[34m172\u001b[39;49;00m,\u001b[34m165\u001b[39;49;00m,\u001b[34m153\u001b[39;49;00m,\u001b[34m142\u001b[39;49;00m,\u001b[34m137\u001b[39;49;00m,\u001b[34m139\u001b[39;49;00m,\u001b[34m148\u001b[39;49;00m,\u001b[34m163\u001b[39;49;00m,\u001b[34m183\u001b[39;49;00m,\u001b[34m204\u001b[39;49;00m,\u001b[34m222\u001b[39;49;00m,\u001b[34m236\u001b[39;49;00m,\u001b[34m246\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m247\u001b[39;49;00m,\u001b[34m234\u001b[39;49;00m,\u001b[34m218\u001b[39;49;00m,\u001b[34m201\u001b[39;49;00m,\u001b[34m186\u001b[39;49;00m,\u001b[34m174\u001b[39;49;00m,\u001b[34m165\u001b[39;49;00m,\u001b[34m157\u001b[39;49;00m,\u001b[34m148\u001b[39;49;00m,\u001b[34m137\u001b[39;49;00m,\u001b[34m130\u001b[39;49;00m,\u001b[34m129\u001b[39;49;00m,\u001b[34m137\u001b[39;49;00m,\u001b[34m151\u001b[39;49;00m,\u001b[34m171\u001b[39;49;00m,\u001b[34m194\u001b[39;49;00m,\u001b[34m214\u001b[39;49;00m,\u001b[34m230\u001b[39;49;00m,\u001b[34m242\u001b[39;49;00m,\u001b[34m248\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m238\u001b[39;49;00m,\u001b[34m222\u001b[39;49;00m,\u001b[34m203\u001b[39;49;00m,\u001b[34m184\u001b[39;49;00m,\u001b[34m168\u001b[39;49;00m,\u001b[34m154\u001b[39;49;00m,\u001b[34m143\u001b[39;49;00m,\u001b[34m132\u001b[39;49;00m,\u001b[34m124\u001b[39;49;00m,\u001b[34m123\u001b[39;49;00m,\u001b[34m130\u001b[39;49;00m,\u001b[34m145\u001b[39;49;00m,\u001b[34m165\u001b[39;49;00m,\u001b[34m188\u001b[39;49;00m,\u001b[34m209\u001b[39;49;00m,\u001b[34m227\u001b[39;49;00m,\u001b[34m239\u001b[39;49;00m,\u001b[34m247\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m244\u001b[39;49;00m,\u001b[34m232\u001b[39;49;00m,\u001b[34m214\u001b[39;49;00m,\u001b[34m194\u001b[39;49;00m,\u001b[34m174\u001b[39;49;00m,\u001b[34m156\u001b[39;49;00m,\u001b[34m142\u001b[39;49;00m,\u001b[34m132\u001b[39;49;00m,\u001b[34m130\u001b[39;49;00m,\u001b[34m134\u001b[39;49;00m,\u001b[34m148\u001b[39;49;00m,\u001b[34m167\u001b[39;49;00m,\u001b[34m189\u001b[39;49;00m,\u001b[34m210\u001b[39;49;00m,\u001b[34m226\u001b[39;49;00m,\u001b[34m238\u001b[39;49;00m,\u001b[34m246\u001b[39;49;00m,\u001b[34m250\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m250\u001b[39;49;00m,\u001b[34m243\u001b[39;49;00m,\u001b[34m231\u001b[39;49;00m,\u001b[34m215\u001b[39;49;00m,\u001b[34m196\u001b[39;49;00m,\u001b[34m178\u001b[39;49;00m,\u001b[34m163\u001b[39;49;00m,\u001b[34m155\u001b[39;49;00m,\u001b[34m156\u001b[39;49;00m,\u001b[34m164\u001b[39;49;00m,\u001b[34m179\u001b[39;49;00m,\u001b[34m197\u001b[39;49;00m,\u001b[34m215\u001b[39;49;00m,\u001b[34m230\u001b[39;49;00m,\u001b[34m240\u001b[39;49;00m,\u001b[34m247\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m246\u001b[39;49;00m,\u001b[34m238\u001b[39;49;00m,\u001b[34m228\u001b[39;49;00m,\u001b[34m217\u001b[39;49;00m,\u001b[34m208\u001b[39;49;00m,\u001b[34m203\u001b[39;49;00m,\u001b[34m204\u001b[39;49;00m,\u001b[34m210\u001b[39;49;00m,\u001b[34m218\u001b[39;49;00m,\u001b[34m228\u001b[39;49;00m,\u001b[34m236\u001b[39;49;00m,\u001b[34m243\u001b[39;49;00m,\u001b[34m248\u001b[39;49;00m,\u001b[34m251\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m245\u001b[39;49;00m,\u001b[34m241\u001b[39;49;00m,\u001b[34m238\u001b[39;49;00m,\u001b[34m237\u001b[39;49;00m,\u001b[34m237\u001b[39;49;00m,\u001b[34m239\u001b[39;49;00m,\u001b[34m242\u001b[39;49;00m,\u001b[34m245\u001b[39;49;00m,\u001b[34m247\u001b[39;49;00m,\u001b[34m250\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m250\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m248\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m249\u001b[39;49;00m,\u001b[34m250\u001b[39;49;00m,\u001b[34m252\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m253\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m254\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m,\u001b[34m255\u001b[39;49;00m])\r\n",
+      "\r\n",
+      "\r\n",
+      "\u001b[34mclass\u001b[39;49;00m \u001b[04m\u001b[32mMnistTransformer\u001b[39;49;00m(\u001b[36mobject\u001b[39;49;00m):\r\n",
+      "\r\n",
+      "    \u001b[34mdef\u001b[39;49;00m \u001b[32m__init__\u001b[39;49;00m(\u001b[36mself\u001b[39;49;00m):\r\n",
+      "        \u001b[34mprint\u001b[39;49;00m(\u001b[33m\"\u001b[39;49;00m\u001b[33minit\u001b[39;49;00m\u001b[33m\"\u001b[39;49;00m);\r\n",
+      "        \r\n",
+      "    \u001b[34mdef\u001b[39;49;00m \u001b[32mpreProcessMNIST\u001b[39;49;00m(\u001b[36mself\u001b[39;49;00m,X):\r\n",
+      "        \u001b[33m'''\u001b[39;49;00m\r\n",
+      "\u001b[33m        Convert values assumed to be in 0-1 range to a value in 0-255.\u001b[39;49;00m\r\n",
+      "\u001b[33m        The remove the training mean needed by the Caffe2 model.\u001b[39;49;00m\r\n",
+      "\u001b[33m        Finally reshape the output to that expected by the model\u001b[39;49;00m\r\n",
+      "\u001b[33m        '''\u001b[39;49;00m\r\n",
+      "        X = X * \u001b[34m255\u001b[39;49;00m\r\n",
+      "        X = \u001b[34m255\u001b[39;49;00m - X\r\n",
+      "        X = (X.reshape(\u001b[34m784\u001b[39;49;00m) - MEANS).reshape(\u001b[34m28\u001b[39;49;00m,\u001b[34m28\u001b[39;49;00m,\u001b[34m1\u001b[39;49;00m)\r\n",
+      "        X = np.transpose(X, (\u001b[34m2\u001b[39;49;00m, \u001b[34m0\u001b[39;49;00m, \u001b[34m1\u001b[39;49;00m))\r\n",
+      "        \u001b[34mreturn\u001b[39;49;00m X\r\n",
+      "\r\n",
+      "    \u001b[34mdef\u001b[39;49;00m \u001b[32mtransform_input\u001b[39;49;00m(\u001b[36mself\u001b[39;49;00m,X,names):\r\n",
+      "        \u001b[34mreturn\u001b[39;49;00m \u001b[36mself\u001b[39;49;00m.preProcessMNIST(X)\r\n",
+      "\r\n",
+      "    \u001b[34mdef\u001b[39;49;00m \u001b[32mtransform_output\u001b[39;49;00m(\u001b[36mself\u001b[39;49;00m,X,names):\r\n",
+      "        \u001b[34mreturn\u001b[39;49;00m X.reshape(\u001b[34m1\u001b[39;49;00m,\u001b[34m10\u001b[39;49;00m)\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!pygmentize MnistTransformer.py"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "---> Installing application source...\n",
+      "Build completed successfully\n"
+     ]
+    }
+   ],
+   "source": [
+    "!s2i build . seldonio/seldon-core-s2i-python3:0.2 mnist-caffe2-transformer:0.1"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 29,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "f2ce59df93344847d070d4218d6bccb5a7d1dab48120967a2e213435a69dce9c\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!docker run --name \"mnist-transformer\" -d --rm -p 5000:5000 mnist-caffe2-transformer:0.1"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Run the Seldon Nvidia Inference Server Proxy Model.\n",
+    "\n",
+    "** CHANGE THE IP ADDRESS BELOW TO THAT OF YOUR RUNNING NVIDIA SERVER **"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 30,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "3bcca7450fadcbb8d54f90af0b40c32f26b875938b742451a4365ced0c5d3ccf\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!docker run --name \"tensorrt-proxy\" -d --rm -p 5001:5001 \\\n",
+    "    -e PREDICTIVE_UNIT_SERVICE_PORT=5001 \\\n",
+    "    -e PREDICTIVE_UNIT_PARAMETERS='[{\"name\":\"url\",\"type\":\"STRING\",\"value\":\"35.204.115.6:8000\"},{\"name\":\"model_name\",\"type\":\"STRING\",\"value\":\"tensorrt_mnist\"},{\"name\":\"protocol\",\"type\":\"STRING\",\"value\":\"HTTP\"}]' \\\n",
+    "    seldonio/nvidia-inference-server-proxy:0.1"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 31,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "TRANSFORMER_URL=\"localhost:5000\"\n",
+    "PREDICTOR_URL=\"localhost:5001\""
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "In the next few steps we illustrate each step by step process and test that out on our running Docker containers."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 32,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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\n",
+      "text/plain": [
+       "<matplotlib.figure.Figure at 0x7ff1285767f0>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "(1, 28, 28)\n"
+     ]
+    }
+   ],
+   "source": [
+    "data = gen_mnist_data(mnist)\n",
+    "response = rest_transform_input_request(TRANSFORMER_URL,data)\n",
+    "transformed = np.array(response['data']['ndarray'])\n",
+    "print(transformed.shape)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 33,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[[[[6.36274109e-08]]\n",
+      "\n",
+      "  [[9.99982357e-01]]\n",
+      "\n",
+      "  [[1.35594092e-07]]\n",
+      "\n",
+      "  [[2.00526298e-08]]\n",
+      "\n",
+      "  [[1.17960089e-05]]\n",
+      "\n",
+      "  [[8.07224509e-08]]\n",
+      "\n",
+      "  [[4.73712625e-08]]\n",
+      "\n",
+      "  [[4.78241873e-06]]\n",
+      "\n",
+      "  [[6.21992911e-07]]\n",
+      "\n",
+      "  [[9.71163061e-08]]]]\n"
+     ]
+    }
+   ],
+   "source": [
+    "response = rest_predict_request(PREDICTOR_URL,transformed)\n",
+    "predictions = np.array(response[\"data\"][\"ndarray\"])\n",
+    "print(predictions)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 34,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "{'data': {'names': None, 'ndarray': [[6.362741089560586e-08, 0.9999823570251465, 1.3559409239860543e-07, 2.005262977888833e-08, 1.1796008948294912e-05, 8.072245094581376e-08, 4.737126246823209e-08, 4.782418727700133e-06, 6.219929105100164e-07, 9.711630610809152e-08]]}}\n"
+     ]
+    }
+   ],
+   "source": [
+    "response = rest_transform_output_request(TRANSFORMER_URL,predictions)\n",
+    "print(response)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 35,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "mnist-transformer\n",
+      "tensorrt-proxy\n"
+     ]
+    }
+   ],
+   "source": [
+    "!docker rm -f mnist-transformer\n",
+    "!docker rm -f tensorrt-proxy"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Test From GCP Cluster"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Set up GCP Kubernetes Cluster\n",
+    "\n",
+    "To run the steps below you will need to:\n",
+    "\n",
+    " 1. Create a GCP Cluster with a GPU node pool with Nvidia V100 GPUs\n",
+    " 2. Enable CUDA on the GPU nodes\n",
+    " 3. Add an Image Pull Secret so you can download the Nvidia Inference Server \n",
+    " \n",
+    " ### Create a GCP Cluster\n",
+    " This can be done from the Google console or via the command line as shown below. Change the cluster name and zones as appropriate for your setup.\n",
+    "\n",
+    "```\n",
+    " gcloud container clusters create myinferenceserver --num-nodes=2 --cluster-version=1.10.6-gke.2\n",
+    " gcloud config set container/cluster myinferenceserver\n",
+    " gcloud container node-pools create gpu-pool --num-nodes=1 --machine-type=n1-standard-8 --accelerator type=nvidia-tesla-v100,count=1\n",
+    " gcloud container clusters get-credentials myinferenceserver\n",
+    "```\n",
+    "\n",
+    "### Enable CUDA on GPU Nodes\n",
+    "\n",
+    "To enable the CUDA drivers on your GPU nodes run:\n",
+    "\n",
+    "```\n",
+    "kubectl apply -f https://raw.githubusercontent.com/GoogleCloudPlatform/container-engine-accelerators/stable/nvidia-driver-installer/cos/daemonset-preloaded.yaml\n",
+    "```\n",
+    "\n",
+    "### Create Image Pull Secret for the Nvidia Repository\n",
+    "\n",
+    " * [Sign up to the NVIDIA GPU Cloud and get an API Key](https://ngc.nvidia.com/)\n",
+    " * Create a kubernetes secret\n",
+    " "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "namespace \"seldon\" created\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl create namespace seldon"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "%env NVIDIA_API_KEY=<enter key here>\n",
+    "%env NVIDIA_CLOUD_EMAIL=<enter email here>"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!kubectl config set-context $(kubectl config current-context) --namespace=seldon"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!kubectl create secret docker-registry ngc \\\n",
+    "    --docker-server=nvcr.io \\\n",
+    "    --docker-username=\\$oauthtoken \\\n",
+    "    --docker-password=${NVIDIA_API_KEY} --docker-email=${NVIDIA_CLOUD_EMAIL}"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Run MNIST Inference Graph"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "serviceaccount \"tiller\" created\n",
+      "clusterrolebinding \"tiller\" created\n",
+      "$HELM_HOME has been configured at /home/clive/.helm.\n",
+      "\n",
+      "Tiller (the Helm server-side component) has been installed into your Kubernetes Cluster.\n",
+      "\n",
+      "Please note: by default, Tiller is deployed with an insecure 'allow unauthenticated users' policy.\n",
+      "To prevent this, run `helm init` with the --tiller-tls-verify flag.\n",
+      "For more information on securing your installation see: https://docs.helm.sh/using_helm/#securing-your-helm-installation\n",
+      "Happy Helming!\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl -n kube-system create sa tiller\n",
+    "!kubectl create clusterrolebinding tiller --clusterrole cluster-admin --serviceaccount=kube-system:tiller\n",
+    "!helm init --service-account tiller"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Waiting for rollout to finish: 0 of 1 updated replicas are available...\n",
+      "deployment \"tiller-deploy\" successfully rolled out\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl rollout status deploy/tiller-deploy -n kube-system"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "NAME:   seldon-core-crd\n",
+      "LAST DEPLOYED: Wed Sep 26 16:52:13 2018\n",
+      "NAMESPACE: seldon\n",
+      "STATUS: DEPLOYED\n",
+      "\n",
+      "RESOURCES:\n",
+      "==> v1beta1/ClusterRoleBinding\n",
+      "NAME                        AGE\n",
+      "seldon-spartakus-volunteer  1s\n",
+      "\n",
+      "==> v1/ConfigMap\n",
+      "NAME                     DATA  AGE\n",
+      "seldon-spartakus-config  3     1s\n",
+      "\n",
+      "==> v1beta1/CustomResourceDefinition\n",
+      "NAME                                         AGE\n",
+      "seldondeployments.machinelearning.seldon.io  1s\n",
+      "\n",
+      "==> v1beta1/Deployment\n",
+      "NAME                        DESIRED  CURRENT  UP-TO-DATE  AVAILABLE  AGE\n",
+      "seldon-spartakus-volunteer  1        0        0           0          1s\n",
+      "\n",
+      "==> v1/ServiceAccount\n",
+      "NAME                        SECRETS  AGE\n",
+      "seldon-spartakus-volunteer  1        1s\n",
+      "\n",
+      "==> v1beta1/ClusterRole\n",
+      "NAME                        AGE\n",
+      "seldon-spartakus-volunteer  1s\n",
+      "\n",
+      "\n",
+      "NOTES:\n",
+      "NOTES: TODO\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "!helm install ../../../helm-charts/seldon-core-crd --name seldon-core-crd \\\n",
+    "    --set usage_metrics.enabled=true"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Context \"gke_kubeflow-testing-213813_europe-west4-a_cluster-4\" modified.\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl config set-context $(kubectl config current-context) --namespace=seldon"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "NAME:   seldon-core\n",
+      "LAST DEPLOYED: Wed Sep 26 16:52:15 2018\n",
+      "NAMESPACE: seldon\n",
+      "STATUS: DEPLOYED\n",
+      "\n",
+      "RESOURCES:\n",
+      "==> v1/RoleBinding\n",
+      "NAME    AGE\n",
+      "seldon  0s\n",
+      "\n",
+      "==> v1beta1/RoleBinding\n",
+      "ambassador  0s\n",
+      "\n",
+      "==> v1beta1/Deployment\n",
+      "NAME                                DESIRED  CURRENT  UP-TO-DATE  AVAILABLE  AGE\n",
+      "seldon-core-ambassador              1        1        1           0          0s\n",
+      "seldon-core-seldon-apiserver        1        1        1           0          0s\n",
+      "seldon-core-seldon-cluster-manager  1        1        1           0          0s\n",
+      "seldon-core-redis                   1        1        1           0          0s\n",
+      "\n",
+      "==> v1/Pod(related)\n",
+      "NAME                                                 READY  STATUS             RESTARTS  AGE\n",
+      "seldon-core-ambassador-778c58bf5d-swkjd              0/2    ContainerCreating  0         0s\n",
+      "seldon-core-seldon-apiserver-6b8dbc978b-cktjb        0/1    ContainerCreating  0         0s\n",
+      "seldon-core-seldon-cluster-manager-596d4674fd-dqbvq  0/1    ContainerCreating  0         0s\n",
+      "seldon-core-redis-8668565565-nkt7q                   0/1    ContainerCreating  0         0s\n",
+      "\n",
+      "==> v1/ServiceAccount\n",
+      "NAME    SECRETS  AGE\n",
+      "seldon  1        0s\n",
+      "\n",
+      "==> v1/ClusterRoleBinding\n",
+      "NAME    AGE\n",
+      "seldon  0s\n",
+      "\n",
+      "==> v1beta1/Role\n",
+      "NAME          AGE\n",
+      "ambassador    0s\n",
+      "seldon-local  0s\n",
+      "\n",
+      "==> v1beta1/ClusterRole\n",
+      "seldon-crd  0s\n",
+      "\n",
+      "==> v1/Service\n",
+      "NAME                          TYPE       CLUSTER-IP     EXTERNAL-IP  PORT(S)                        AGE\n",
+      "seldon-core-ambassador-admin  NodePort   10.43.245.152  <none>       8877:31931/TCP                 0s\n",
+      "seldon-core-ambassador        NodePort   10.43.254.206  <none>       8080:32597/TCP                 0s\n",
+      "seldon-core-seldon-apiserver  NodePort   10.43.243.99   <none>       8080:30835/TCP,5000:31146/TCP  0s\n",
+      "seldon-core-redis             ClusterIP  10.43.240.11   <none>       6379/TCP                       0s\n",
+      "\n",
+      "\n",
+      "NOTES:\n",
+      "Thank you for installing Seldon Core.\n",
+      "\n",
+      "Documentation can be found at https://github.com/SeldonIO/seldon-core\n",
+      "\n",
+      "\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "!helm install ../../../helm-charts/seldon-core --name seldon-core \\\n",
+    "        --namespace seldon \\\n",
+    "        --set ambassador.enabled=true"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Waiting for rollout to finish: 0 of 1 updated replicas are available...\n",
+      "deployment \"seldon-core-seldon-cluster-manager\" successfully rolled out\n",
+      "deployment \"seldon-core-seldon-apiserver\" successfully rolled out\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl rollout status deploy/seldon-core-seldon-cluster-manager -n seldon\n",
+    "!kubectl rollout status deploy/seldon-core-seldon-apiserver -n seldon"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "NAME:   running-wolf\n",
+      "LAST DEPLOYED: Wed Sep 26 17:08:44 2018\n",
+      "NAMESPACE: seldon\n",
+      "STATUS: DEPLOYED\n",
+      "\n",
+      "RESOURCES:\n",
+      "==> v1alpha2/SeldonDeployment\n",
+      "NAME          AGE\n",
+      "nvidia-mnist  1s\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "!helm install nvidia-mnist --namespace seldon --set tfserving.model_base_path=${MODEL_REPOSITORY_BUCKET}"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!helm template nvidia-mnist --name nvidia-mnist --namespace seldon --set tfserving.model_base_path=${MODEL_REPOSITORY_BUCKET} > mnist.json"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!sed '1,2d' mnist.json > tmp.json"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/svg+xml": [
+       "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\n",
+       "<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n",
+       " \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n",
+       "<!-- Generated by graphviz version 2.38.0 (20140413.2041)\n",
+       " -->\n",
+       "<!-- Title: %3 Pages: 1 -->\n",
+       "<svg width=\"208pt\" height=\"243pt\"\n",
+       " viewBox=\"0.00 0.00 208.00 243.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n",
+       "<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 239)\">\n",
+       "<title>%3</title>\n",
+       "<polygon fill=\"white\" stroke=\"none\" points=\"-4,4 -4,-239 204,-239 204,4 -4,4\"/>\n",
+       "<g id=\"clust1\" class=\"cluster\"><title>cluster_0</title>\n",
+       "<polygon fill=\"none\" stroke=\"black\" points=\"8,-8 8,-227 192,-227 192,-8 8,-8\"/>\n",
+       "<text text-anchor=\"middle\" x=\"100\" y=\"-211.8\" font-family=\"Times,serif\" font-size=\"14.00\">predictor&#45;0</text>\n",
+       "</g>\n",
+       "<!-- mnist&#45;transformer0 -->\n",
+       "<g id=\"node1\" class=\"node\"><title>mnist&#45;transformer0</title>\n",
+       "<polygon fill=\"none\" stroke=\"black\" points=\"158,-196 42,-196 42,-160 158,-160 158,-196\"/>\n",
+       "<text text-anchor=\"middle\" x=\"100\" y=\"-174.3\" font-family=\"Times,serif\" font-size=\"14.00\">mnist&#45;transformer</text>\n",
+       "</g>\n",
+       "<!-- mnist&#45;transformer0endpoint -->\n",
+       "<g id=\"node2\" class=\"node\"><title>mnist&#45;transformer0endpoint</title>\n",
+       "<ellipse fill=\"none\" stroke=\"black\" cx=\"48\" cy=\"-106\" rx=\"32.4942\" ry=\"18\"/>\n",
+       "<text text-anchor=\"middle\" x=\"48\" y=\"-102.3\" font-family=\"Times,serif\" font-size=\"14.00\">REST</text>\n",
+       "</g>\n",
+       "<!-- mnist&#45;transformer0&#45;&gt;mnist&#45;transformer0endpoint -->\n",
+       "<g id=\"edge1\" class=\"edge\"><title>mnist&#45;transformer0&#45;&gt;mnist&#45;transformer0endpoint</title>\n",
+       "<path fill=\"none\" stroke=\"black\" d=\"M87.1461,-159.697C80.7907,-151.141 73.0133,-140.672 66.0702,-131.325\"/>\n",
+       "<polygon fill=\"black\" stroke=\"black\" points=\"68.766,-129.085 59.9931,-123.145 63.1468,-133.259 68.766,-129.085\"/>\n",
+       "</g>\n",
+       "<!-- nvidia&#45;proxy -->\n",
+       "<g id=\"node3\" class=\"node\"><title>nvidia&#45;proxy</title>\n",
+       "<polygon fill=\"none\" stroke=\"black\" points=\"184,-124 98,-124 98,-88 184,-88 184,-124\"/>\n",
+       "<text text-anchor=\"middle\" x=\"141\" y=\"-102.3\" font-family=\"Times,serif\" font-size=\"14.00\">nvidia&#45;proxy</text>\n",
+       "</g>\n",
+       "<!-- mnist&#45;transformer0&#45;&gt;nvidia&#45;proxy -->\n",
+       "<g id=\"edge3\" class=\"edge\"><title>mnist&#45;transformer0&#45;&gt;nvidia&#45;proxy</title>\n",
+       "<path fill=\"none\" stroke=\"black\" d=\"M110.135,-159.697C114.852,-151.644 120.562,-141.894 125.782,-132.982\"/>\n",
+       "<polygon fill=\"black\" stroke=\"black\" points=\"128.948,-134.502 130.982,-124.104 122.908,-130.964 128.948,-134.502\"/>\n",
+       "</g>\n",
+       "<!-- nvidia&#45;proxyendpoint -->\n",
+       "<g id=\"node4\" class=\"node\"><title>nvidia&#45;proxyendpoint</title>\n",
+       "<ellipse fill=\"none\" stroke=\"black\" cx=\"141\" cy=\"-34\" rx=\"32.4942\" ry=\"18\"/>\n",
+       "<text text-anchor=\"middle\" x=\"141\" y=\"-30.3\" font-family=\"Times,serif\" font-size=\"14.00\">REST</text>\n",
+       "</g>\n",
+       "<!-- nvidia&#45;proxy&#45;&gt;nvidia&#45;proxyendpoint -->\n",
+       "<g id=\"edge2\" class=\"edge\"><title>nvidia&#45;proxy&#45;&gt;nvidia&#45;proxyendpoint</title>\n",
+       "<path fill=\"none\" stroke=\"black\" d=\"M141,-87.6966C141,-79.9827 141,-70.7125 141,-62.1124\"/>\n",
+       "<polygon fill=\"black\" stroke=\"black\" points=\"144.5,-62.1043 141,-52.1043 137.5,-62.1044 144.5,-62.1043\"/>\n",
+       "</g>\n",
+       "</g>\n",
+       "</svg>\n"
+      ],
+      "text/plain": [
+       "<graphviz.dot.Digraph at 0x7f1e48ccbac8>"
+      ]
+     },
+     "execution_count": 23,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "get_graph(\"tmp.json\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\u001b[04m\u001b[31;01m-\u001b[39;49;00m\u001b[04m\u001b[31;01m-\u001b[39;49;00m\u001b[04m\u001b[31;01m-\u001b[39;49;00m\r\n",
+      "\u001b[04m\u001b[31;01m#\u001b[39;49;00m \u001b[04m\u001b[31;01mS\u001b[39;49;00m\u001b[04m\u001b[31;01mo\u001b[39;49;00m\u001b[04m\u001b[31;01mu\u001b[39;49;00m\u001b[04m\u001b[31;01mr\u001b[39;49;00m\u001b[04m\u001b[31;01mc\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01m:\u001b[39;49;00m \u001b[04m\u001b[31;01mn\u001b[39;49;00m\u001b[04m\u001b[31;01mv\u001b[39;49;00m\u001b[04m\u001b[31;01mi\u001b[39;49;00m\u001b[04m\u001b[31;01md\u001b[39;49;00m\u001b[04m\u001b[31;01mi\u001b[39;49;00m\u001b[04m\u001b[31;01ma\u001b[39;49;00m\u001b[04m\u001b[31;01m-\u001b[39;49;00m\u001b[04m\u001b[31;01mm\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\u001b[04m\u001b[31;01mi\u001b[39;49;00m\u001b[04m\u001b[31;01ms\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01m/\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01mm\u001b[39;49;00m\u001b[04m\u001b[31;01mp\u001b[39;49;00m\u001b[04m\u001b[31;01ml\u001b[39;49;00m\u001b[04m\u001b[31;01ma\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01ms\u001b[39;49;00m\u001b[04m\u001b[31;01m/\u001b[39;49;00m\u001b[04m\u001b[31;01mm\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\u001b[04m\u001b[31;01mi\u001b[39;49;00m\u001b[04m\u001b[31;01ms\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01m_\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\u001b[04m\u001b[31;01mv\u001b[39;49;00m\u001b[04m\u001b[31;01mi\u001b[39;49;00m\u001b[04m\u001b[31;01md\u001b[39;49;00m\u001b[04m\u001b[31;01mi\u001b[39;49;00m\u001b[04m\u001b[31;01ma\u001b[39;49;00m\u001b[04m\u001b[31;01m_\u001b[39;49;00m\u001b[04m\u001b[31;01md\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01mp\u001b[39;49;00m\u001b[04m\u001b[31;01ml\u001b[39;49;00m\u001b[04m\u001b[31;01mo\u001b[39;49;00m\u001b[04m\u001b[31;01my\u001b[39;49;00m\u001b[04m\u001b[31;01mm\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01m.\u001b[39;49;00m\u001b[04m\u001b[31;01mj\u001b[39;49;00m\u001b[04m\u001b[31;01ms\u001b[39;49;00m\u001b[04m\u001b[31;01mo\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\r\n",
+      "{\r\n",
+      "    \u001b[34;01m\"apiVersion\"\u001b[39;49;00m: \u001b[33m\"machinelearning.seldon.io/v1alpha2\"\u001b[39;49;00m,\r\n",
+      "    \u001b[34;01m\"kind\"\u001b[39;49;00m: \u001b[33m\"SeldonDeployment\"\u001b[39;49;00m,\r\n",
+      "    \u001b[34;01m\"metadata\"\u001b[39;49;00m: {\r\n",
+      "        \u001b[34;01m\"labels\"\u001b[39;49;00m: {\r\n",
+      "            \u001b[34;01m\"app\"\u001b[39;49;00m: \u001b[33m\"seldon\"\u001b[39;49;00m\r\n",
+      "        },\r\n",
+      "        \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"nvidia-mnist\"\u001b[39;49;00m,\r\n",
+      "\t\u001b[34;01m\"namespace\"\u001b[39;49;00m: \u001b[33m\"seldon\"\u001b[39;49;00m\r\n",
+      "    },\r\n",
+      "    \u001b[34;01m\"spec\"\u001b[39;49;00m: {\r\n",
+      "        \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"caffe2-mnist\"\u001b[39;49;00m,\r\n",
+      "        \u001b[34;01m\"predictors\"\u001b[39;49;00m: [\r\n",
+      "            {\r\n",
+      "                \u001b[34;01m\"componentSpecs\"\u001b[39;49;00m: [{\r\n",
+      "                    \u001b[34;01m\"spec\"\u001b[39;49;00m: {\r\n",
+      "                        \u001b[34;01m\"containers\"\u001b[39;49;00m: [\r\n",
+      "                            {\r\n",
+      "                                \u001b[34;01m\"image\"\u001b[39;49;00m: \u001b[33m\"seldonio/mnist-caffe2-transformer:0.1\"\u001b[39;49;00m,\r\n",
+      "                                \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"mnist-transformer\"\u001b[39;49;00m\r\n",
+      "                            },\r\n",
+      "                            {\r\n",
+      "                                \u001b[34;01m\"image\"\u001b[39;49;00m: \u001b[33m\"seldonio/nvidia-inference-server-proxy:0.1\"\u001b[39;49;00m,\r\n",
+      "                                \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"nvidia-proxy\"\u001b[39;49;00m\r\n",
+      "                            },\r\n",
+      "\t\t\t    {\r\n",
+      "\t\t\t\t\u001b[34;01m\"args\"\u001b[39;49;00m: [\r\n",
+      "\t\t\t\t    \u001b[33m\"--model-store=gs://seldon-inference-server-model-store\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t],\r\n",
+      "\t\t\t\t\u001b[34;01m\"command\"\u001b[39;49;00m: [\r\n",
+      "\t\t\t\t    \u001b[33m\"inference_server\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t],\r\n",
+      "\t\t\t\t\u001b[34;01m\"image\"\u001b[39;49;00m: \u001b[33m\"nvcr.io/nvidia/inferenceserver:18.08.1-py2\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\u001b[34;01m\"livenessProbe\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t    \u001b[34;01m\"failureThreshold\"\u001b[39;49;00m: \u001b[34m3\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"handler\"\u001b[39;49;00m:{\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"httpGet\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t\t    \u001b[34;01m\"path\"\u001b[39;49;00m: \u001b[33m\"/api/health/live\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t    \u001b[34;01m\"port\"\u001b[39;49;00m: \u001b[34m8000\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t    \u001b[34;01m\"scheme\"\u001b[39;49;00m: \u001b[33m\"HTTP\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t\t}\r\n",
+      "\t\t\t\t    },\r\n",
+      "\t\t\t\t    \u001b[34;01m\"initialDelaySeconds\"\u001b[39;49;00m: \u001b[34m5\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"periodSeconds\"\u001b[39;49;00m: \u001b[34m5\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"successThreshold\"\u001b[39;49;00m: \u001b[34m1\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"timeoutSeconds\"\u001b[39;49;00m: \u001b[34m1\u001b[39;49;00m\r\n",
+      "\t\t\t\t},\r\n",
+      "\t\t\t\t\u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"inference-server\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\u001b[34;01m\"ports\"\u001b[39;49;00m: [\r\n",
+      "\t\t\t\t    {\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"containerPort\"\u001b[39;49;00m: \u001b[34m8000\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"protocol\"\u001b[39;49;00m: \u001b[33m\"TCP\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t    },\r\n",
+      "\t\t\t\t    {\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"containerPort\"\u001b[39;49;00m: \u001b[34m8001\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"protocol\"\u001b[39;49;00m: \u001b[33m\"TCP\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t    },\r\n",
+      "\t\t\t\t    {\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"containerPort\"\u001b[39;49;00m: \u001b[34m8002\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"protocol\"\u001b[39;49;00m: \u001b[33m\"TCP\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t    }\r\n",
+      "\t\t\t\t],\r\n",
+      "\t\t\t\t\u001b[34;01m\"readinessProbe\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t    \u001b[34;01m\"failureThreshold\"\u001b[39;49;00m: \u001b[34m3\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"handler\"\u001b[39;49;00m:{\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"httpGet\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t\t    \u001b[34;01m\"path\"\u001b[39;49;00m: \u001b[33m\"/api/health/ready\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t    \u001b[34;01m\"port\"\u001b[39;49;00m: \u001b[34m8000\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t    \u001b[34;01m\"scheme\"\u001b[39;49;00m: \u001b[33m\"HTTP\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t\t}\r\n",
+      "\t\t\t\t    },\r\n",
+      "\t\t\t\t    \u001b[34;01m\"initialDelaySeconds\"\u001b[39;49;00m: \u001b[34m5\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"periodSeconds\"\u001b[39;49;00m: \u001b[34m5\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"successThreshold\"\u001b[39;49;00m: \u001b[34m1\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"timeoutSeconds\"\u001b[39;49;00m: \u001b[34m1\u001b[39;49;00m\r\n",
+      "\t\t\t\t},\r\n",
+      "\t\t\t\t\u001b[34;01m\"resources\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t    \u001b[34;01m\"limits\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"nvidia.com/gpu\"\u001b[39;49;00m: \u001b[33m\"1\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t    },\r\n",
+      "\t\t\t\t    \u001b[34;01m\"requests\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"cpu\"\u001b[39;49;00m: \u001b[33m\"100m\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"nvidia.com/gpu\"\u001b[39;49;00m: \u001b[33m\"1\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t    }\r\n",
+      "\t\t\t\t},\r\n",
+      "\t\t\t\t\u001b[34;01m\"securityContext\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t    \u001b[34;01m\"runAsUser\"\u001b[39;49;00m: \u001b[34m1000\u001b[39;49;00m\r\n",
+      "\t\t\t\t}\r\n",
+      "\t\t\t    }\r\n",
+      "\t\t\t],\r\n",
+      "\t\t\t\u001b[34;01m\"terminationGracePeriodSeconds\"\u001b[39;49;00m: \u001b[34m1\u001b[39;49;00m,\r\n",
+      "\t\t\t\u001b[34;01m\"imagePullSecrets\"\u001b[39;49;00m: [\r\n",
+      "\t\t\t    {\r\n",
+      "\t\t\t\t\u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"ngc\"\u001b[39;49;00m\r\n",
+      "\t\t\t    }\r\n",
+      "\t\t\t]\r\n",
+      "\t\t    }\r\n",
+      "\t\t}],\r\n",
+      "                \u001b[34;01m\"graph\"\u001b[39;49;00m: {\r\n",
+      "                    \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"mnist-transformer\"\u001b[39;49;00m,\r\n",
+      "                    \u001b[34;01m\"endpoint\"\u001b[39;49;00m: { \u001b[34;01m\"type\"\u001b[39;49;00m : \u001b[33m\"REST\"\u001b[39;49;00m },\r\n",
+      "                    \u001b[34;01m\"type\"\u001b[39;49;00m: \u001b[33m\"TRANSFORMER\"\u001b[39;49;00m,\r\n",
+      "\t\t    \u001b[34;01m\"children\"\u001b[39;49;00m: [\r\n",
+      "\t\t\t{\r\n",
+      "\t\t\t    \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"nvidia-proxy\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"endpoint\"\u001b[39;49;00m: { \u001b[34;01m\"type\"\u001b[39;49;00m : \u001b[33m\"REST\"\u001b[39;49;00m },\r\n",
+      "\t\t\t    \u001b[34;01m\"type\"\u001b[39;49;00m: \u001b[33m\"MODEL\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"children\"\u001b[39;49;00m: [],\r\n",
+      "\t\t\t    \u001b[34;01m\"parameters\"\u001b[39;49;00m:\r\n",
+      "\t\t\t    [\r\n",
+      "\t\t\t\t{\r\n",
+      "\t\t\t\t    \u001b[34;01m\"name\"\u001b[39;49;00m:\u001b[33m\"url\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"type\"\u001b[39;49;00m:\u001b[33m\"STRING\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"value\"\u001b[39;49;00m:\u001b[33m\"127.0.0.1:8000\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t},\r\n",
+      "\t\t\t\t{\r\n",
+      "\t\t\t\t    \u001b[34;01m\"name\"\u001b[39;49;00m:\u001b[33m\"model_name\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"type\"\u001b[39;49;00m:\u001b[33m\"STRING\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"value\"\u001b[39;49;00m:\u001b[33m\"tensorrt_mnist\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t},\r\n",
+      "\t\t\t\t{\r\n",
+      "\t\t\t\t    \u001b[34;01m\"name\"\u001b[39;49;00m:\u001b[33m\"protocol\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"type\"\u001b[39;49;00m:\u001b[33m\"STRING\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[34;01m\"value\"\u001b[39;49;00m:\u001b[33m\"HTTP\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t}\r\n",
+      "\t\t\t    ]\r\n",
+      "\t\t\t}\r\n",
+      "\t\t    ]\r\n",
+      "                },\r\n",
+      "                \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"mnist-nvidia\"\u001b[39;49;00m,\r\n",
+      "                \u001b[34;01m\"replicas\"\u001b[39;49;00m: \u001b[34m1\u001b[39;49;00m\r\n",
+      "            }\r\n",
+      "        ]\r\n",
+      "    }\r\n",
+      "}\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!pygmentize mnist.json"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "**Port forward Ambassador**\n",
+    "\n",
+    "```\n",
+    "kubectl port-forward $(kubectl get pods -n seldon -l service=ambassador -o jsonpath='{.items[0].metadata.name}') -n seldon 8003:8080\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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\n",
+      "text/plain": [
+       "<matplotlib.figure.Figure at 0x7f1e48ccb470>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "200\n",
+      "{\n",
+      "  \"meta\": {\n",
+      "    \"puid\": \"rnaaqs28vus3vsi3t0n19q3240\",\n",
+      "    \"tags\": {\n",
+      "    },\n",
+      "    \"routing\": {\n",
+      "      \"mnist-transformer\": -1\n",
+      "    }\n",
+      "  },\n",
+      "  \"data\": {\n",
+      "    \"names\": [\"t:0\", \"t:1\", \"t:2\", \"t:3\", \"t:4\", \"t:5\", \"t:6\", \"t:7\", \"t:8\", \"t:9\"],\n",
+      "    \"tensor\": {\n",
+      "      \"shape\": [1, 10, 1, 1],\n",
+      "      \"values\": [3.5578280637360224E-14, 6.920142844174526E-12, 4.660422221747229E-12, 4.3726320849736544E-14, 1.0, 3.065522011583277E-12, 4.4698250499129386E-14, 3.898779521449569E-9, 9.478232001342013E-13, 3.0998212330324293E-10]\n",
+      "    }\n",
+      "  }\n",
+      "}\n"
+     ]
+    }
+   ],
+   "source": [
+    "data = gen_mnist_data(mnist)\n",
+    "data = data.reshape((784))\n",
+    "rest_request_ambassador(\"nvidia-mnist\",endpoint=\"localhost:8003\",arr=data)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Analytics and Load Test"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "NAME:   seldon-core-analytics\n",
+      "LAST DEPLOYED: Mon Sep 24 14:40:46 2018\n",
+      "NAMESPACE: seldon\n",
+      "STATUS: DEPLOYED\n",
+      "\n",
+      "RESOURCES:\n",
+      "==> v1beta1/ClusterRoleBinding\n",
+      "NAME        AGE\n",
+      "prometheus  0s\n",
+      "\n",
+      "==> v1/Job\n",
+      "NAME                            DESIRED  SUCCESSFUL  AGE\n",
+      "grafana-prom-import-dashboards  1        0           0s\n",
+      "\n",
+      "==> v1/Service\n",
+      "NAME                      TYPE       CLUSTER-IP     EXTERNAL-IP  PORT(S)       AGE\n",
+      "alertmanager              ClusterIP  10.39.243.208  <none>       80/TCP        0s\n",
+      "grafana-prom              NodePort   10.39.246.121  <none>       80:30588/TCP  0s\n",
+      "prometheus-node-exporter  ClusterIP  None           <none>       9100/TCP      0s\n",
+      "prometheus-seldon         ClusterIP  10.39.253.10   <none>       80/TCP        0s\n",
+      "\n",
+      "==> v1beta1/DaemonSet\n",
+      "NAME                      DESIRED  CURRENT  READY  UP-TO-DATE  AVAILABLE  NODE SELECTOR  AGE\n",
+      "prometheus-node-exporter  1        1        0      1           0          <none>         0s\n",
+      "\n",
+      "==> v1/ConfigMap\n",
+      "NAME                       DATA  AGE\n",
+      "alertmanager-server-conf   1     0s\n",
+      "grafana-import-dashboards  7     0s\n",
+      "prometheus-rules           4     0s\n",
+      "prometheus-server-conf     1     0s\n",
+      "\n",
+      "==> v1beta1/ClusterRole\n",
+      "NAME        AGE\n",
+      "prometheus  0s\n",
+      "\n",
+      "==> v1beta1/Deployment\n",
+      "NAME                     DESIRED  CURRENT  UP-TO-DATE  AVAILABLE  AGE\n",
+      "alertmanager-deployment  1        1        1           0          0s\n",
+      "grafana-prom-deployment  1        1        1           0          0s\n",
+      "prometheus-deployment    1        1        1           0          0s\n",
+      "\n",
+      "==> v1/Pod(related)\n",
+      "NAME                                      READY  STATUS             RESTARTS  AGE\n",
+      "grafana-prom-import-dashboards-wrtnb      0/1    ContainerCreating  0         0s\n",
+      "alertmanager-deployment-557b99ccf8-ztdnk  0/1    ContainerCreating  0         0s\n",
+      "grafana-prom-deployment-dd84b7788-zwwls   0/1    ContainerCreating  0         0s\n",
+      "prometheus-node-exporter-zg9cg            0/1    ContainerCreating  0         0s\n",
+      "prometheus-deployment-78dd89b44f-8ntcw    0/1    Pending            0         0s\n",
+      "\n",
+      "==> v1/Secret\n",
+      "NAME                 TYPE    DATA  AGE\n",
+      "grafana-prom-secret  Opaque  1     0s\n",
+      "\n",
+      "==> v1/ServiceAccount\n",
+      "NAME        SECRETS  AGE\n",
+      "prometheus  1        0s\n",
+      "\n",
+      "\n",
+      "NOTES:\n",
+      "NOTES: TODO\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "!helm install ../../../helm-charts/seldon-core-analytics --name seldon-core-analytics \\\n",
+    "    --set grafana_prom_admin_password=password \\\n",
+    "    --set persistence.enabled=false \\\n",
+    "    --namespace seldon"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "node \"gke-cluster-1-default-pool-5e22f4e3-djd7\" labeled\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl label nodes $(kubectl get nodes -o jsonpath='{.items[0].metadata.name}') role=locust"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "NAME:   loadtest\n",
+      "LAST DEPLOYED: Mon Sep 24 14:40:53 2018\n",
+      "NAMESPACE: seldon\n",
+      "STATUS: DEPLOYED\n",
+      "\n",
+      "RESOURCES:\n",
+      "==> v1/ReplicationController\n",
+      "NAME             DESIRED  CURRENT  READY  AGE\n",
+      "locust-slave-1   1        1        0      0s\n",
+      "locust-master-1  1        1        0      0s\n",
+      "\n",
+      "==> v1/Service\n",
+      "NAME             TYPE      CLUSTER-IP    EXTERNAL-IP  PORT(S)                                       AGE\n",
+      "locust-master-1  NodePort  10.39.254.17  <none>       5557:31880/TCP,5558:30185/TCP,8089:32570/TCP  0s\n",
+      "\n",
+      "==> v1/Pod(related)\n",
+      "NAME                   READY  STATUS             RESTARTS  AGE\n",
+      "locust-slave-1-p5n6f   0/1    ContainerCreating  0         0s\n",
+      "locust-master-1-ksk7g  0/1    ContainerCreating  0         0s\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "!helm install seldon-core-loadtesting --name loadtest  \\\n",
+    "    --namespace seldon \\\n",
+    "    --repo https://storage.googleapis.com/seldon-charts \\\n",
+    "    --set locust.script=mnist_rest_locust.py \\\n",
+    "    --set locust.host=http://caffe2-mnist:8000 \\\n",
+    "    --set oauth.enabled=false \\\n",
+    "    --set locust.hatchRate=1 \\\n",
+    "    --set locust.clients=1 \\\n",
+    "    --set loadtest.sendFeedback=1 \\\n",
+    "    --set locust.minWait=0 \\\n",
+    "    --set locust.maxWait=0 \\\n",
+    "    --set replicaCount=1 \\\n",
+    "    --set data.size=784\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "You should port-foward the grafana dashboard\n",
+    "\n",
+    "```\n",
+    "kubectl port-forward $(kubectl get pods -n seldon -l app=grafana-prom-server -o jsonpath='{.items[0].metadata.name}') -n seldon 3000:3000\n",
+    "```\n",
+    "\n",
+    "You can then view an analytics dashboard inside the cluster at http://localhost:3000/dashboard/db/prediction-analytics?refresh=5s&orgId=1. Your IP address may be different. get it via minikube ip. Login with:\n",
+    "\n",
+    "    Username : admin\n",
+    "\n",
+    "    password : password (as set when starting seldon-core-analytics above)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.6.4"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}
diff --git a/examples/models/nvidia-mnist/tensorrt_mnist/config.pbtxt b/examples/models/nvidia-mnist/tensorrt_mnist/config.pbtxt
new file mode 100644
index 0000000000..57032e9d08
--- /dev/null
+++ b/examples/models/nvidia-mnist/tensorrt_mnist/config.pbtxt
@@ -0,0 +1,19 @@
+name: "tensorrt_mnist"
+platform: "tensorrt_plan"
+max_batch_size: 1
+input [
+  {
+    name: "data"
+    data_type: TYPE_FP32
+    format: FORMAT_NCHW
+    dims: [ 1, 28, 28 ]
+  }
+]
+output [
+  {
+    name: "prob"
+    data_type: TYPE_FP32
+    dims: [ 10, 1, 1 ]
+    label_filename: "mnist_labels.txt"
+  }
+]
diff --git a/examples/models/nvidia-mnist/tensorrt_mnist/mnist_labels.txt b/examples/models/nvidia-mnist/tensorrt_mnist/mnist_labels.txt
new file mode 100644
index 0000000000..8b1acc12b6
--- /dev/null
+++ b/examples/models/nvidia-mnist/tensorrt_mnist/mnist_labels.txt
@@ -0,0 +1,10 @@
+0
+1
+2
+3
+4
+5
+6
+7
+8
+9
diff --git a/examples/models/tfserving-mnist/Makefile b/examples/models/tfserving-mnist/Makefile
new file mode 100644
index 0000000000..7241b4f57b
--- /dev/null
+++ b/examples/models/tfserving-mnist/Makefile
@@ -0,0 +1,8 @@
+
+
+clean:
+	rm -rf MNIST_data
+	rm -rf mnist-model/1
+	rm -rf serving
+	rm -f tmp.json
+	rm -f mnist.json
diff --git a/examples/models/tfserving-mnist/mnist_tfserving_deployment.json.template b/examples/models/tfserving-mnist/mnist_tfserving_deployment.json.template
new file mode 100644
index 0000000000..73693ef039
--- /dev/null
+++ b/examples/models/tfserving-mnist/mnist_tfserving_deployment.json.template
@@ -0,0 +1,93 @@
+{
+    "apiVersion": "machinelearning.seldon.io/v1alpha2",
+    "kind": "SeldonDeployment",
+    "metadata": {
+        "labels": {
+            "app": "seldon"
+        },
+        "name": "tfserving-mnist"
+    },
+    "spec": {
+        "name": "tf-mnist",
+        "predictors": [
+            {
+                "componentSpecs": [{
+                    "spec": {
+                        "containers": [
+                            {
+                                "image": "seldonio/tfserving-proxy:0.1",
+                                "name": "tfserving-proxy"
+                            },
+			    {
+				"args": [
+				    "/usr/bin/tensorflow_model_server",
+				    "--port=8000",
+				    "--model_name=mnist-model",
+				    "--model_base_path=gs://seldon-tfserving-store/mnist-model"
+				],
+				"image": "gcr.io/kubeflow-images-public/tensorflow-serving-1.7:v20180604-0da89b8a",
+				"name": "mnist-model",
+				"ports": [
+				    {
+					"containerPort": 8000,
+					"protocol": "TCP"
+				    }
+				],
+				"resources": {
+				    "limits": {
+					"cpu": "4",
+					"memory": "4Gi"
+				    },
+				    "requests": {
+					"cpu": "1",
+					"memory": "1Gi"
+				    }
+				},
+				"securityContext": {
+				    "runAsUser": 1000
+				}
+			    }
+			],
+			"terminationGracePeriodSeconds": 1
+		    }
+		}],
+                "graph": {
+		    "name": "tfserving-proxy",
+		    "endpoint": { "type" : "REST" },
+		    "type": "MODEL",
+		    "children": [],
+		    "parameters":
+		    [
+			{
+			    "name":"grpc_endpoint",
+			    "type":"STRING",
+			    "value":"localhost:8000"
+			},
+			{
+			    "name":"model_name",
+			    "type":"STRING",
+			    "value":"mnist-model"
+			},
+			{
+			    "name":"model_output",
+			    "type":"STRING",
+			    "value":"scores"
+			},
+			{
+			    "name":"model_input",
+			    "type":"STRING",
+			    "value":"images"
+			},
+			{
+			    "name":"signature_name",
+			    "type":"STRING",
+			    "value":"predict_images"
+			}
+		    ]
+		},
+                "name": "mnist-tfserving",
+                "replicas": 1
+            }
+        ]
+    }
+}
diff --git a/examples/models/tfserving-mnist/tfserving-mnist.ipynb b/examples/models/tfserving-mnist/tfserving-mnist.ipynb
new file mode 100644
index 0000000000..68f1eab8ad
--- /dev/null
+++ b/examples/models/tfserving-mnist/tfserving-mnist.ipynb
@@ -0,0 +1,960 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Seldon and TensorFlow Serving MNIST Example\n",
+    "\n",
+    "This example shows how you can combine Seldon with Tensorflo Serving. We will use a Seldon Tensorflow Serving proxy model image that will forward Seldon internal microservice prediction calls out to a Tensorflow serving server.\n",
+    "\n",
+    "The example will use the MNIST digit classification task with the example MNIST model.\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Setup"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/home/clive/anaconda3/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
+      "  from ._conv import register_converters as _register_converters\n"
+     ]
+    }
+   ],
+   "source": [
+    "%matplotlib inline\n",
+    "import requests\n",
+    "from random import randint,random\n",
+    "import json\n",
+    "from matplotlib import pyplot as plt\n",
+    "import numpy as np\n",
+    "from tensorflow.examples.tutorials.mnist import input_data\n",
+    "import sys\n",
+    "sys.path.append(\"../../../notebooks\")\n",
+    "from visualizer import get_graph"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def gen_image(arr):\n",
+    "    two_d = (np.reshape(arr, (28, 28)) * 255).astype(np.uint8)\n",
+    "    plt.imshow(two_d,cmap=plt.cm.gray_r, interpolation='nearest')\n",
+    "    return plt\n",
+    "\n",
+    "def download_mnist():\n",
+    "    return input_data.read_data_sets(\"MNIST_data/\", one_hot = True)\n",
+    "\n",
+    "def rest_predict_request(endpoint,data):\n",
+    "    request = {\"data\":{\"ndarray\":data.tolist()}}\n",
+    "    response = requests.post(\n",
+    "                \"http://\"+endpoint+\"/predict\",\n",
+    "                data={\"json\":json.dumps(request),\"isDefault\":True})\n",
+    "    return response.json()   \n",
+    "\n",
+    "def rest_transform_input_request(endpoint,data):\n",
+    "    request = {\"data\":{\"ndarray\":data.tolist()}}\n",
+    "    response = requests.post(\n",
+    "                \"http://\"+endpoint+\"/transform-input\",\n",
+    "                data={\"json\":json.dumps(request),\"isDefault\":True})\n",
+    "    return response.json()   \n",
+    "\n",
+    "def rest_transform_output_request(endpoint,data):\n",
+    "    request = {\"data\":{\"ndarray\":data.tolist()}}\n",
+    "    response = requests.post(\n",
+    "                \"http://\"+endpoint+\"/transform-output\",\n",
+    "                data={\"json\":json.dumps(request),\"isDefault\":True})\n",
+    "    return response.json()   \n",
+    "\n",
+    "def rest_request_ambassador(deploymentName,endpoint=\"localhost:8003\",arr=None):\n",
+    "    payload = {\"data\":{\"names\":[\"a\",\"b\"],\"tensor\":{\"shape\":[1,784],\"values\":arr.tolist()}}}\n",
+    "    response = requests.post(\n",
+    "                \"http://\"+endpoint+\"/seldon/\"+deploymentName+\"/api/v0.1/predictions\",\n",
+    "                json=payload)\n",
+    "    print(response.status_code)\n",
+    "    print(response.text)\n",
+    "\n",
+    "\n",
+    "def gen_mnist_data(mnist):\n",
+    "    batch_xs, batch_ys = mnist.train.next_batch(1)\n",
+    "    chosen=0\n",
+    "    gen_image(batch_xs[chosen]).show()\n",
+    "    data = batch_xs[chosen].reshape((1,784))\n",
+    "    return data\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "WARNING:tensorflow:From <ipython-input-2-8029beac92e7>:7: read_data_sets (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use alternatives such as official/mnist/dataset.py from tensorflow/models.\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:260: maybe_download (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please write your own downloading logic.\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/base.py:252: _internal_retry.<locals>.wrap.<locals>.wrapped_fn (from tensorflow.contrib.learn.python.learn.datasets.base) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use urllib or similar directly.\n",
+      "Successfully downloaded train-images-idx3-ubyte.gz 9912422 bytes.\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:262: extract_images (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use tf.data to implement this functionality.\n",
+      "Extracting MNIST_data/train-images-idx3-ubyte.gz\n",
+      "Successfully downloaded train-labels-idx1-ubyte.gz 28881 bytes.\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:267: extract_labels (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use tf.data to implement this functionality.\n",
+      "Extracting MNIST_data/train-labels-idx1-ubyte.gz\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:110: dense_to_one_hot (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use tf.one_hot on tensors.\n",
+      "Successfully downloaded t10k-images-idx3-ubyte.gz 1648877 bytes.\n",
+      "Extracting MNIST_data/t10k-images-idx3-ubyte.gz\n",
+      "Successfully downloaded t10k-labels-idx1-ubyte.gz 4542 bytes.\n",
+      "Extracting MNIST_data/t10k-labels-idx1-ubyte.gz\n",
+      "WARNING:tensorflow:From /home/clive/anaconda3/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/datasets/mnist.py:290: DataSet.__init__ (from tensorflow.contrib.learn.python.learn.datasets.mnist) is deprecated and will be removed in a future version.\n",
+      "Instructions for updating:\n",
+      "Please use alternatives such as official/mnist/dataset.py from tensorflow/models.\n"
+     ]
+    }
+   ],
+   "source": [
+    "mnist = download_mnist()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Create MNIST Model Repository\n",
+    "You will need tensorflow installed to run these steps."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Train Tensorflow MNIST example model"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Cloning into 'serving'...\n",
+      "remote: Enumerating objects: 75, done.\u001b[K\n",
+      "remote: Counting objects: 100% (75/75), done.\u001b[K\n",
+      "remote: Compressing objects: 100% (43/43), done.\u001b[K\n",
+      "remote: Total 11251 (delta 53), reused 49 (delta 32), pack-reused 11176\u001b[K\n",
+      "Receiving objects: 100% (11251/11251), 3.62 MiB | 3.02 MiB/s, done.\n",
+      "Resolving deltas: 100% (8109/8109), done.\n",
+      "Checking connectivity... done.\n"
+     ]
+    }
+   ],
+   "source": [
+    "!git clone https://github.com/tensorflow/serving.git"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "/home/clive/anaconda3/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
+      "  from ._conv import register_converters as _register_converters\n",
+      "Training model...\n",
+      "Successfully downloaded train-images-idx3-ubyte.gz 9912422 bytes.\n",
+      "Extracting /tmp/train-images-idx3-ubyte.gz\n",
+      "Successfully downloaded train-labels-idx1-ubyte.gz 28881 bytes.\n",
+      "Extracting /tmp/train-labels-idx1-ubyte.gz\n",
+      "Successfully downloaded t10k-images-idx3-ubyte.gz 1648877 bytes.\n",
+      "Extracting /tmp/t10k-images-idx3-ubyte.gz\n",
+      "Successfully downloaded t10k-labels-idx1-ubyte.gz 4542 bytes.\n",
+      "Extracting /tmp/t10k-labels-idx1-ubyte.gz\n",
+      "2018-09-26 15:48:35.512962: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA\n",
+      "training accuracy 0.9162\n",
+      "Done training!\n",
+      "Exporting trained model to b'../../../mnist-model/1'\n",
+      "Done exporting!\n"
+     ]
+    }
+   ],
+   "source": [
+    "!cd serving/tensorflow_serving/example && python mnist_saved_model.py --training_iteration=100000 ../../../mnist-model"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Copy Model to Google Bucket"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "env: MODEL_REPOSITORY_BUCKET=gs://seldon-tfserving-store\n"
+     ]
+    }
+   ],
+   "source": [
+    "# CHANGE THIS TO YOUR OWN CHOSEN GOOGLE BUCKET NAME\n",
+    "%env MODEL_REPOSITORY_BUCKET=gs://seldon-tfserving-store"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Creating gs://seldon-tfserving-store/...\n",
+      "ServiceException: 409 Bucket seldon-tfserving-store already exists.\n"
+     ]
+    }
+   ],
+   "source": [
+    "!gsutil mb ${MODEL_REPOSITORY_BUCKET}"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Copying file://mnist-model/1/saved_model.pb [Content-Type=application/octet-stream]...\n",
+      "Copying file://mnist-model/1/variables/variables.data-00000-of-00001 [Content-Type=application/octet-stream]...\n",
+      "Copying file://mnist-model/1/variables/variables.index [Content-Type=application/octet-stream]...\n",
+      "- [3 files][ 48.6 KiB/ 48.6 KiB]                                                \n",
+      "Operation completed over 3 objects/48.6 KiB.                                     \n"
+     ]
+    }
+   ],
+   "source": [
+    "!gsutil cp -r mnist-model ${MODEL_REPOSITORY_BUCKET}"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Test From GCP Cluster"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Run MNIST Inference Graph"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "serviceaccount \"tiller\" created\n",
+      "clusterrolebinding \"tiller\" created\n",
+      "$HELM_HOME has been configured at /home/clive/.helm.\n",
+      "\n",
+      "Tiller (the Helm server-side component) has been installed into your Kubernetes Cluster.\n",
+      "\n",
+      "Please note: by default, Tiller is deployed with an insecure 'allow unauthenticated users' policy.\n",
+      "To prevent this, run `helm init` with the --tiller-tls-verify flag.\n",
+      "For more information on securing your installation see: https://docs.helm.sh/using_helm/#securing-your-helm-installation\n",
+      "Happy Helming!\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl -n kube-system create sa tiller\n",
+    "!kubectl create clusterrolebinding tiller --clusterrole cluster-admin --serviceaccount=kube-system:tiller\n",
+    "!helm init --service-account tiller"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "deployment \"tiller-deploy\" successfully rolled out\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl rollout status deploy/tiller-deploy -n kube-system"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Error: a release named seldon-core-crd already exists.\r\n",
+      "Run: helm ls --all seldon-core-crd; to check the status of the release\r\n",
+      "Or run: helm del --purge seldon-core-crd; to delete it\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!helm install ../../../helm-charts/seldon-core-crd --name seldon-core-crd \\\n",
+    "    --set usage_metrics.enabled=true"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "namespace \"seldon\" created\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl create namespace seldon"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Context \"gke_kubeflow-testing-213813_europe-west4-a_cluster-2\" modified.\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl config set-context $(kubectl config current-context) --namespace=seldon"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "NAME:   seldon-core\n",
+      "LAST DEPLOYED: Wed Sep 26 15:53:10 2018\n",
+      "NAMESPACE: seldon\n",
+      "STATUS: DEPLOYED\n",
+      "\n",
+      "RESOURCES:\n",
+      "==> v1/ClusterRoleBinding\n",
+      "NAME    AGE\n",
+      "seldon  1s\n",
+      "\n",
+      "==> v1beta1/RoleBinding\n",
+      "NAME        AGE\n",
+      "ambassador  1s\n",
+      "\n",
+      "==> v1beta1/Deployment\n",
+      "NAME                                DESIRED  CURRENT  UP-TO-DATE  AVAILABLE  AGE\n",
+      "seldon-core-ambassador              1        1        1           0          0s\n",
+      "seldon-core-seldon-apiserver        1        1        1           0          0s\n",
+      "seldon-core-seldon-cluster-manager  1        1        1           0          0s\n",
+      "seldon-core-redis                   1        1        1           0          0s\n",
+      "\n",
+      "==> v1/ServiceAccount\n",
+      "NAME    SECRETS  AGE\n",
+      "seldon  1        1s\n",
+      "\n",
+      "==> v1beta1/ClusterRole\n",
+      "NAME        AGE\n",
+      "seldon-crd  1s\n",
+      "\n",
+      "==> v1beta1/Role\n",
+      "ambassador    1s\n",
+      "seldon-local  1s\n",
+      "\n",
+      "==> v1/RoleBinding\n",
+      "NAME    AGE\n",
+      "seldon  1s\n",
+      "\n",
+      "==> v1/Service\n",
+      "NAME                          TYPE       CLUSTER-IP     EXTERNAL-IP  PORT(S)                        AGE\n",
+      "seldon-core-ambassador-admin  NodePort   10.19.242.55   <none>       8877:32623/TCP                 1s\n",
+      "seldon-core-ambassador        NodePort   10.19.252.46   <none>       8080:32423/TCP                 1s\n",
+      "seldon-core-seldon-apiserver  NodePort   10.19.240.170  <none>       8080:31470/TCP,5000:30245/TCP  1s\n",
+      "seldon-core-redis             ClusterIP  10.19.248.179  <none>       6379/TCP                       1s\n",
+      "\n",
+      "==> v1/Pod(related)\n",
+      "NAME                                                 READY  STATUS             RESTARTS  AGE\n",
+      "seldon-core-ambassador-778c58bf5d-tk9hz              0/2    ContainerCreating  0         0s\n",
+      "seldon-core-seldon-apiserver-6b8dbc978b-56std        0/1    ContainerCreating  0         0s\n",
+      "seldon-core-seldon-cluster-manager-596d4674fd-swqs6  0/1    ContainerCreating  0         0s\n",
+      "seldon-core-redis-8668565565-5g98j                   0/1    ContainerCreating  0         0s\n",
+      "\n",
+      "\n",
+      "NOTES:\n",
+      "Thank you for installing Seldon Core.\n",
+      "\n",
+      "Documentation can be found at https://github.com/SeldonIO/seldon-core\n",
+      "\n",
+      "\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "!helm install ../../../helm-charts/seldon-core --name seldon-core \\\n",
+    "        --namespace seldon \\\n",
+    "        --set ambassador.enabled=true"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Waiting for rollout to finish: 0 of 1 updated replicas are available...\n",
+      "deployment \"seldon-core-seldon-cluster-manager\" successfully rolled out\n",
+      "Waiting for rollout to finish: 0 of 1 updated replicas are available...\n",
+      "deployment \"seldon-core-seldon-apiserver\" successfully rolled out\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl rollout status deploy/seldon-core-seldon-cluster-manager -n seldon\n",
+    "!kubectl rollout status deploy/seldon-core-seldon-apiserver -n seldon"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "NAME:   tfserving-mnist\n",
+      "LAST DEPLOYED: Wed Sep 26 15:56:41 2018\n",
+      "NAMESPACE: seldon\n",
+      "STATUS: DEPLOYED\n",
+      "\n",
+      "RESOURCES:\n",
+      "==> v1alpha2/SeldonDeployment\n",
+      "NAME             AGE\n",
+      "tfserving-mnist  0s\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "!helm install tfserving-mnist --name tfserving-mnist --namespace seldon --set tfserving.model_base_path=${MODEL_REPOSITORY_BUCKET}/mnist-model "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 19,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!helm template tfserving-mnist --name tfserving-mnist --namespace seldon --set tfserving.model_base_path=${MODEL_REPOSITORY_BUCKET}/mnist-model > mnist.json"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 20,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!sed '1,2d' mnist.json > tmp.json"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/svg+xml": [
+       "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\n",
+       "<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\n",
+       " \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\n",
+       "<!-- Generated by graphviz version 2.38.0 (20140413.2041)\n",
+       " -->\n",
+       "<!-- Title: %3 Pages: 1 -->\n",
+       "<svg width=\"142pt\" height=\"171pt\"\n",
+       " viewBox=\"0.00 0.00 142.00 171.00\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n",
+       "<g id=\"graph0\" class=\"graph\" transform=\"scale(1 1) rotate(0) translate(4 167)\">\n",
+       "<title>%3</title>\n",
+       "<polygon fill=\"white\" stroke=\"none\" points=\"-4,4 -4,-167 138,-167 138,4 -4,4\"/>\n",
+       "<g id=\"clust1\" class=\"cluster\"><title>cluster_0</title>\n",
+       "<polygon fill=\"none\" stroke=\"black\" points=\"8,-8 8,-155 126,-155 126,-8 8,-8\"/>\n",
+       "<text text-anchor=\"middle\" x=\"67\" y=\"-139.8\" font-family=\"Times,serif\" font-size=\"14.00\">predictor&#45;0</text>\n",
+       "</g>\n",
+       "<!-- tfserving&#45;proxy0 -->\n",
+       "<g id=\"node1\" class=\"node\"><title>tfserving&#45;proxy0</title>\n",
+       "<polygon fill=\"none\" stroke=\"black\" points=\"117.5,-124 16.5,-124 16.5,-88 117.5,-88 117.5,-124\"/>\n",
+       "<text text-anchor=\"middle\" x=\"67\" y=\"-102.3\" font-family=\"Times,serif\" font-size=\"14.00\">tfserving&#45;proxy</text>\n",
+       "</g>\n",
+       "<!-- tfserving&#45;proxy0endpoint -->\n",
+       "<g id=\"node2\" class=\"node\"><title>tfserving&#45;proxy0endpoint</title>\n",
+       "<ellipse fill=\"none\" stroke=\"black\" cx=\"67\" cy=\"-34\" rx=\"32.4942\" ry=\"18\"/>\n",
+       "<text text-anchor=\"middle\" x=\"67\" y=\"-30.3\" font-family=\"Times,serif\" font-size=\"14.00\">REST</text>\n",
+       "</g>\n",
+       "<!-- tfserving&#45;proxy0&#45;&gt;tfserving&#45;proxy0endpoint -->\n",
+       "<g id=\"edge1\" class=\"edge\"><title>tfserving&#45;proxy0&#45;&gt;tfserving&#45;proxy0endpoint</title>\n",
+       "<path fill=\"none\" stroke=\"black\" d=\"M67,-87.6966C67,-79.9827 67,-70.7125 67,-62.1124\"/>\n",
+       "<polygon fill=\"black\" stroke=\"black\" points=\"70.5001,-62.1043 67,-52.1043 63.5001,-62.1044 70.5001,-62.1043\"/>\n",
+       "</g>\n",
+       "</g>\n",
+       "</svg>\n"
+      ],
+      "text/plain": [
+       "<graphviz.dot.Digraph at 0x7f7485a2da58>"
+      ]
+     },
+     "execution_count": 21,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "get_graph(\"tmp.json\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\u001b[04m\u001b[31;01m-\u001b[39;49;00m\u001b[04m\u001b[31;01m-\u001b[39;49;00m\u001b[04m\u001b[31;01m-\u001b[39;49;00m\r\n",
+      "\u001b[04m\u001b[31;01m#\u001b[39;49;00m \u001b[04m\u001b[31;01mS\u001b[39;49;00m\u001b[04m\u001b[31;01mo\u001b[39;49;00m\u001b[04m\u001b[31;01mu\u001b[39;49;00m\u001b[04m\u001b[31;01mr\u001b[39;49;00m\u001b[04m\u001b[31;01mc\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01m:\u001b[39;49;00m \u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01mf\u001b[39;49;00m\u001b[04m\u001b[31;01ms\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01mr\u001b[39;49;00m\u001b[04m\u001b[31;01mv\u001b[39;49;00m\u001b[04m\u001b[31;01mi\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\u001b[04m\u001b[31;01mg\u001b[39;49;00m\u001b[04m\u001b[31;01m-\u001b[39;49;00m\u001b[04m\u001b[31;01mm\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\u001b[04m\u001b[31;01mi\u001b[39;49;00m\u001b[04m\u001b[31;01ms\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01m/\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01mm\u001b[39;49;00m\u001b[04m\u001b[31;01mp\u001b[39;49;00m\u001b[04m\u001b[31;01ml\u001b[39;49;00m\u001b[04m\u001b[31;01ma\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01ms\u001b[39;49;00m\u001b[04m\u001b[31;01m/\u001b[39;49;00m\u001b[04m\u001b[31;01mm\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\u001b[04m\u001b[31;01mi\u001b[39;49;00m\u001b[04m\u001b[31;01ms\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01m_\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01mf\u001b[39;49;00m\u001b[04m\u001b[31;01ms\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01mr\u001b[39;49;00m\u001b[04m\u001b[31;01mv\u001b[39;49;00m\u001b[04m\u001b[31;01mi\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\u001b[04m\u001b[31;01mg\u001b[39;49;00m\u001b[04m\u001b[31;01m_\u001b[39;49;00m\u001b[04m\u001b[31;01md\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01mp\u001b[39;49;00m\u001b[04m\u001b[31;01ml\u001b[39;49;00m\u001b[04m\u001b[31;01mo\u001b[39;49;00m\u001b[04m\u001b[31;01my\u001b[39;49;00m\u001b[04m\u001b[31;01mm\u001b[39;49;00m\u001b[04m\u001b[31;01me\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\u001b[04m\u001b[31;01mt\u001b[39;49;00m\u001b[04m\u001b[31;01m.\u001b[39;49;00m\u001b[04m\u001b[31;01mj\u001b[39;49;00m\u001b[04m\u001b[31;01ms\u001b[39;49;00m\u001b[04m\u001b[31;01mo\u001b[39;49;00m\u001b[04m\u001b[31;01mn\u001b[39;49;00m\r\n",
+      "{\r\n",
+      "    \u001b[34;01m\"apiVersion\"\u001b[39;49;00m: \u001b[33m\"machinelearning.seldon.io/v1alpha2\"\u001b[39;49;00m,\r\n",
+      "    \u001b[34;01m\"kind\"\u001b[39;49;00m: \u001b[33m\"SeldonDeployment\"\u001b[39;49;00m,\r\n",
+      "    \u001b[34;01m\"metadata\"\u001b[39;49;00m: {\r\n",
+      "        \u001b[34;01m\"labels\"\u001b[39;49;00m: {\r\n",
+      "            \u001b[34;01m\"app\"\u001b[39;49;00m: \u001b[33m\"seldon\"\u001b[39;49;00m\r\n",
+      "        },\r\n",
+      "        \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"tfserving-mnist\"\u001b[39;49;00m,\r\n",
+      "\t\u001b[34;01m\"namespace\"\u001b[39;49;00m: \u001b[33m\"seldon\"\u001b[39;49;00m\t\r\n",
+      "    },\r\n",
+      "    \u001b[34;01m\"spec\"\u001b[39;49;00m: {\r\n",
+      "        \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"tf-mnist\"\u001b[39;49;00m,\r\n",
+      "        \u001b[34;01m\"predictors\"\u001b[39;49;00m: [\r\n",
+      "            {\r\n",
+      "                \u001b[34;01m\"componentSpecs\"\u001b[39;49;00m: [{\r\n",
+      "                    \u001b[34;01m\"spec\"\u001b[39;49;00m: {\r\n",
+      "                        \u001b[34;01m\"containers\"\u001b[39;49;00m: [\r\n",
+      "                            {\r\n",
+      "                                \u001b[34;01m\"image\"\u001b[39;49;00m: \u001b[33m\"seldonio/tfserving-proxy:0.1\"\u001b[39;49;00m,\r\n",
+      "                                \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"tfserving-proxy\"\u001b[39;49;00m\r\n",
+      "                            },\r\n",
+      "\t\t\t    {\r\n",
+      "\t\t\t\t\u001b[34;01m\"args\"\u001b[39;49;00m: [\r\n",
+      "\t\t\t\t    \u001b[33m\"/usr/bin/tensorflow_model_server\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[33m\"--port=8000\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[33m\"--model_name=mnist-model\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t    \u001b[33m\"--model_base_path=gs://seldon-tfserving-store/mnist-model\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t],\r\n",
+      "\t\t\t\t\u001b[34;01m\"image\"\u001b[39;49;00m: \u001b[33m\"gcr.io/kubeflow-images-public/tensorflow-serving-1.7:v20180604-0da89b8a\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"mnist-model\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\u001b[34;01m\"ports\"\u001b[39;49;00m: [\r\n",
+      "\t\t\t\t    {\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"containerPort\"\u001b[39;49;00m: \u001b[34m8000\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"protocol\"\u001b[39;49;00m: \u001b[33m\"TCP\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t    }\r\n",
+      "\t\t\t\t],\r\n",
+      "\t\t\t\t\u001b[34;01m\"resources\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t    \u001b[34;01m\"limits\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"cpu\"\u001b[39;49;00m: \u001b[33m\"4\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"memory\"\u001b[39;49;00m: \u001b[33m\"4Gi\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t    },\r\n",
+      "\t\t\t\t    \u001b[34;01m\"requests\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"cpu\"\u001b[39;49;00m: \u001b[33m\"1\"\u001b[39;49;00m,\r\n",
+      "\t\t\t\t\t\u001b[34;01m\"memory\"\u001b[39;49;00m: \u001b[33m\"1Gi\"\u001b[39;49;00m\r\n",
+      "\t\t\t\t    }\r\n",
+      "\t\t\t\t},\r\n",
+      "\t\t\t\t\u001b[34;01m\"securityContext\"\u001b[39;49;00m: {\r\n",
+      "\t\t\t\t    \u001b[34;01m\"runAsUser\"\u001b[39;49;00m: \u001b[34m1000\u001b[39;49;00m\r\n",
+      "\t\t\t\t}\r\n",
+      "\t\t\t    }\r\n",
+      "\t\t\t],\r\n",
+      "\t\t\t\u001b[34;01m\"terminationGracePeriodSeconds\"\u001b[39;49;00m: \u001b[34m1\u001b[39;49;00m\r\n",
+      "\t\t    }\r\n",
+      "\t\t}],\r\n",
+      "                \u001b[34;01m\"graph\"\u001b[39;49;00m: {\r\n",
+      "\t\t    \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"tfserving-proxy\"\u001b[39;49;00m,\r\n",
+      "\t\t    \u001b[34;01m\"endpoint\"\u001b[39;49;00m: { \u001b[34;01m\"type\"\u001b[39;49;00m : \u001b[33m\"REST\"\u001b[39;49;00m },\r\n",
+      "\t\t    \u001b[34;01m\"type\"\u001b[39;49;00m: \u001b[33m\"MODEL\"\u001b[39;49;00m,\r\n",
+      "\t\t    \u001b[34;01m\"children\"\u001b[39;49;00m: [],\r\n",
+      "\t\t    \u001b[34;01m\"parameters\"\u001b[39;49;00m:\r\n",
+      "\t\t    [\r\n",
+      "\t\t\t{\r\n",
+      "\t\t\t    \u001b[34;01m\"name\"\u001b[39;49;00m:\u001b[33m\"grpc_endpoint\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"type\"\u001b[39;49;00m:\u001b[33m\"STRING\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"value\"\u001b[39;49;00m:\u001b[33m\"localhost:8000\"\u001b[39;49;00m\r\n",
+      "\t\t\t},\r\n",
+      "\t\t\t{\r\n",
+      "\t\t\t    \u001b[34;01m\"name\"\u001b[39;49;00m:\u001b[33m\"model_name\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"type\"\u001b[39;49;00m:\u001b[33m\"STRING\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"value\"\u001b[39;49;00m:\u001b[33m\"mnist-model\"\u001b[39;49;00m\r\n",
+      "\t\t\t},\r\n",
+      "\t\t\t{\r\n",
+      "\t\t\t    \u001b[34;01m\"name\"\u001b[39;49;00m:\u001b[33m\"model_output\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"type\"\u001b[39;49;00m:\u001b[33m\"STRING\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"value\"\u001b[39;49;00m:\u001b[33m\"scores\"\u001b[39;49;00m\r\n",
+      "\t\t\t},\r\n",
+      "\t\t\t{\r\n",
+      "\t\t\t    \u001b[34;01m\"name\"\u001b[39;49;00m:\u001b[33m\"model_input\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"type\"\u001b[39;49;00m:\u001b[33m\"STRING\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"value\"\u001b[39;49;00m:\u001b[33m\"images\"\u001b[39;49;00m\r\n",
+      "\t\t\t},\r\n",
+      "\t\t\t{\r\n",
+      "\t\t\t    \u001b[34;01m\"name\"\u001b[39;49;00m:\u001b[33m\"signature_name\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"type\"\u001b[39;49;00m:\u001b[33m\"STRING\"\u001b[39;49;00m,\r\n",
+      "\t\t\t    \u001b[34;01m\"value\"\u001b[39;49;00m:\u001b[33m\"predict_images\"\u001b[39;49;00m\r\n",
+      "\t\t\t}\r\n",
+      "\t\t    ]\r\n",
+      "\t\t},\r\n",
+      "                \u001b[34;01m\"name\"\u001b[39;49;00m: \u001b[33m\"mnist-tfserving\"\u001b[39;49;00m,\r\n",
+      "                \u001b[34;01m\"replicas\"\u001b[39;49;00m: \u001b[34m1\u001b[39;49;00m\r\n",
+      "            }\r\n",
+      "        ]\r\n",
+      "    }\r\n",
+      "}\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!pygmentize mnist.json"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "**Port forward Ambassador**\n",
+    "\n",
+    "```\n",
+    "kubectl port-forward $(kubectl get pods -n seldon -l service=ambassador -o jsonpath='{.items[0].metadata.name}') -n seldon 8003:8080\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "image/png": "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\n",
+      "text/plain": [
+       "<matplotlib.figure.Figure at 0x7f7484af5e48>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "200\n",
+      "{\n",
+      "  \"meta\": {\n",
+      "    \"puid\": \"faun2l1ftmp3r90cm6s5vnjo8a\",\n",
+      "    \"tags\": {\n",
+      "    },\n",
+      "    \"routing\": {\n",
+      "    }\n",
+      "  },\n",
+      "  \"data\": {\n",
+      "    \"names\": [\"t:0\", \"t:1\", \"t:2\", \"t:3\", \"t:4\", \"t:5\", \"t:6\", \"t:7\", \"t:8\", \"t:9\"],\n",
+      "    \"tensor\": {\n",
+      "      \"shape\": [1, 10],\n",
+      "      \"values\": [1.124886839534156E-4, 9.059235132891086E-12, 0.0028196186758577824, 0.005111460108309984, 2.414313166809734E-5, 0.6415375471115112, 1.74979015810095E-6, 1.2785285052885342E-9, 0.34967318177223206, 7.198737002909184E-4]\n",
+      "    }\n",
+      "  }\n",
+      "}\n"
+     ]
+    }
+   ],
+   "source": [
+    "data = gen_mnist_data(mnist)\n",
+    "data = data.reshape((784))\n",
+    "rest_request_ambassador(\"tfserving-mnist\",endpoint=\"localhost:8003\",arr=data)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Analytics and Load Test"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "NAME:   seldon-core-analytics\n",
+      "LAST DEPLOYED: Mon Sep 24 15:52:34 2018\n",
+      "NAMESPACE: seldon\n",
+      "STATUS: DEPLOYED\n",
+      "\n",
+      "RESOURCES:\n",
+      "==> v1/Service\n",
+      "NAME                      TYPE       CLUSTER-IP     EXTERNAL-IP  PORT(S)       AGE\n",
+      "alertmanager              ClusterIP  10.19.253.159  <none>       80/TCP        1s\n",
+      "grafana-prom              NodePort   10.19.251.82   <none>       80:30690/TCP  1s\n",
+      "prometheus-node-exporter  ClusterIP  None           <none>       9100/TCP      1s\n",
+      "prometheus-seldon         ClusterIP  10.19.245.167  <none>       80/TCP        1s\n",
+      "\n",
+      "==> v1/Secret\n",
+      "NAME                 TYPE    DATA  AGE\n",
+      "grafana-prom-secret  Opaque  1     2s\n",
+      "\n",
+      "==> v1/ServiceAccount\n",
+      "NAME        SECRETS  AGE\n",
+      "prometheus  1        2s\n",
+      "\n",
+      "==> v1/Job\n",
+      "NAME                            DESIRED  SUCCESSFUL  AGE\n",
+      "grafana-prom-import-dashboards  1        0           1s\n",
+      "\n",
+      "==> v1beta1/Deployment\n",
+      "NAME                     DESIRED  CURRENT  UP-TO-DATE  AVAILABLE  AGE\n",
+      "alertmanager-deployment  1        1        1           0          1s\n",
+      "grafana-prom-deployment  1        1        1           0          1s\n",
+      "prometheus-deployment    1        1        1           0          1s\n",
+      "\n",
+      "==> v1beta1/DaemonSet\n",
+      "NAME                      DESIRED  CURRENT  READY  UP-TO-DATE  AVAILABLE  NODE SELECTOR  AGE\n",
+      "prometheus-node-exporter  1        1        0      1           0          <none>         1s\n",
+      "\n",
+      "==> v1/Pod(related)\n",
+      "NAME                                      READY  STATUS             RESTARTS  AGE\n",
+      "grafana-prom-import-dashboards-wzl55      0/1    ContainerCreating  0         1s\n",
+      "alertmanager-deployment-557b99ccf8-m27mm  0/1    ContainerCreating  0         1s\n",
+      "grafana-prom-deployment-dd84b7788-gmdkc   0/1    ContainerCreating  0         1s\n",
+      "prometheus-node-exporter-pnv9k            0/1    ContainerCreating  0         1s\n",
+      "prometheus-deployment-78dd89b44f-7t62q    0/1    ContainerCreating  0         1s\n",
+      "\n",
+      "==> v1/ConfigMap\n",
+      "NAME                       DATA  AGE\n",
+      "alertmanager-server-conf   1     2s\n",
+      "grafana-import-dashboards  7     2s\n",
+      "prometheus-rules           4     2s\n",
+      "prometheus-server-conf     1     2s\n",
+      "\n",
+      "==> v1beta1/ClusterRole\n",
+      "NAME        AGE\n",
+      "prometheus  2s\n",
+      "\n",
+      "==> v1beta1/ClusterRoleBinding\n",
+      "NAME        AGE\n",
+      "prometheus  2s\n",
+      "\n",
+      "\n",
+      "NOTES:\n",
+      "NOTES: TODO\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "!helm install ../../../helm-charts/seldon-core-analytics --name seldon-core-analytics \\\n",
+    "    --set grafana_prom_admin_password=password \\\n",
+    "    --set persistence.enabled=false \\\n",
+    "    --namespace seldon"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "node \"gke-cluster-2-default-pool-54fbe785-wmcb\" labeled\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "!kubectl label nodes $(kubectl get nodes -o jsonpath='{.items[0].metadata.name}') role=locust"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "NAME:   loadtest\n",
+      "LAST DEPLOYED: Mon Sep 24 15:52:43 2018\n",
+      "NAMESPACE: seldon\n",
+      "STATUS: DEPLOYED\n",
+      "\n",
+      "RESOURCES:\n",
+      "==> v1/ReplicationController\n",
+      "NAME             DESIRED  CURRENT  READY  AGE\n",
+      "locust-slave-1   1        1        0      0s\n",
+      "locust-master-1  1        1        0      0s\n",
+      "\n",
+      "==> v1/Service\n",
+      "NAME             TYPE      CLUSTER-IP     EXTERNAL-IP  PORT(S)                                       AGE\n",
+      "locust-master-1  NodePort  10.19.255.204  <none>       5557:32342/TCP,5558:30493/TCP,8089:32034/TCP  0s\n",
+      "\n",
+      "==> v1/Pod(related)\n",
+      "NAME                   READY  STATUS             RESTARTS  AGE\n",
+      "locust-slave-1-zncg4   0/1    ContainerCreating  0         0s\n",
+      "locust-master-1-9z8gm  0/1    ContainerCreating  0         0s\n",
+      "\n",
+      "\n"
+     ]
+    }
+   ],
+   "source": [
+    "!helm install seldon-core-loadtesting --name loadtest  \\\n",
+    "    --namespace seldon \\\n",
+    "    --repo https://storage.googleapis.com/seldon-charts \\\n",
+    "    --set locust.script=mnist_rest_locust.py \\\n",
+    "    --set locust.host=http://tf-mnist:8000 \\\n",
+    "    --set oauth.enabled=false \\\n",
+    "    --set locust.hatchRate=1 \\\n",
+    "    --set locust.clients=1 \\\n",
+    "    --set loadtest.sendFeedback=1 \\\n",
+    "    --set locust.minWait=0 \\\n",
+    "    --set locust.maxWait=0 \\\n",
+    "    --set replicaCount=1 \\\n",
+    "    --set data.size=784\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "You should port-foward the grafana dashboard\n",
+    "\n",
+    "```\n",
+    "kubectl port-forward $(kubectl get pods -n seldon -l app=grafana-prom-server -o jsonpath='{.items[0].metadata.name}') -n seldon 3000:3000\n",
+    "```\n",
+    "\n",
+    "You can then view an analytics dashboard inside the cluster at http://localhost:3000/dashboard/db/prediction-analytics?refresh=5s&orgId=1. Your IP address may be different. get it via minikube ip. Login with:\n",
+    "\n",
+    "    Username : admin\n",
+    "\n",
+    "    password : password (as set when starting seldon-core-analytics above)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.6.4"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}
diff --git a/examples/models/tfserving-mnist/tfserving-mnist/Chart.yaml b/examples/models/tfserving-mnist/tfserving-mnist/Chart.yaml
new file mode 100644
index 0000000000..e746d40542
--- /dev/null
+++ b/examples/models/tfserving-mnist/tfserving-mnist/Chart.yaml
@@ -0,0 +1,6 @@
+apiVersion: v1
+description: Seldon MNIST TFServing Example
+name: tfserving-mnist
+sources:
+- https://github.com/SeldonIO/seldon-core
+version: 0.1
diff --git a/examples/models/tfserving-mnist/tfserving-mnist/templates/mnist_tfserving_deployment.json b/examples/models/tfserving-mnist/tfserving-mnist/templates/mnist_tfserving_deployment.json
new file mode 100644
index 0000000000..6e4637c3ff
--- /dev/null
+++ b/examples/models/tfserving-mnist/tfserving-mnist/templates/mnist_tfserving_deployment.json
@@ -0,0 +1,94 @@
+{
+    "apiVersion": "machinelearning.seldon.io/v1alpha2",
+    "kind": "SeldonDeployment",
+    "metadata": {
+        "labels": {
+            "app": "seldon"
+        },
+        "name": "tfserving-mnist",
+	"namespace": "{{ .Release.Namespace }}"	
+    },
+    "spec": {
+        "name": "tf-mnist",
+        "predictors": [
+            {
+                "componentSpecs": [{
+                    "spec": {
+                        "containers": [
+                            {
+                                "image": "seldonio/tfserving-proxy:0.1",
+                                "name": "tfserving-proxy"
+                            },
+			    {
+				"args": [
+				    "/usr/bin/tensorflow_model_server",
+				    "--port={{ .Values.tfserving.port }}",
+				    "--model_name=mnist-model",
+				    "--model_base_path={{ .Values.tfserving.model_base_path }}"
+				],
+				"image": "gcr.io/kubeflow-images-public/tensorflow-serving-1.7:v20180604-0da89b8a",
+				"name": "mnist-model",
+				"ports": [
+				    {
+					"containerPort": {{ .Values.tfserving.port }},
+					"protocol": "TCP"
+				    }
+				],
+				"resources": {
+				    "limits": {
+					"cpu": "4",
+					"memory": "4Gi"
+				    },
+				    "requests": {
+					"cpu": "1",
+					"memory": "1Gi"
+				    }
+				},
+				"securityContext": {
+				    "runAsUser": 1000
+				}
+			    }
+			],
+			"terminationGracePeriodSeconds": 1
+		    }
+		}],
+                "graph": {
+		    "name": "tfserving-proxy",
+		    "endpoint": { "type" : "REST" },
+		    "type": "MODEL",
+		    "children": [],
+		    "parameters":
+		    [
+			{
+			    "name":"grpc_endpoint",
+			    "type":"STRING",
+			    "value":"localhost:{{ .Values.tfserving.port }}"
+			},
+			{
+			    "name":"model_name",
+			    "type":"STRING",
+			    "value":"mnist-model"
+			},
+			{
+			    "name":"model_output",
+			    "type":"STRING",
+			    "value":"scores"
+			},
+			{
+			    "name":"model_input",
+			    "type":"STRING",
+			    "value":"images"
+			},
+			{
+			    "name":"signature_name",
+			    "type":"STRING",
+			    "value":"predict_images"
+			}
+		    ]
+		},
+                "name": "mnist-tfserving",
+                "replicas": 1
+            }
+        ]
+    }
+}
diff --git a/examples/models/tfserving-mnist/tfserving-mnist/values.yaml b/examples/models/tfserving-mnist/tfserving-mnist/values.yaml
new file mode 100644
index 0000000000..2049eb89ec
--- /dev/null
+++ b/examples/models/tfserving-mnist/tfserving-mnist/values.yaml
@@ -0,0 +1,4 @@
+tfserving:
+  model_base_path: gs://seldon-tfserving-store/mnist-model
+  port: 8000
+  
diff --git a/examples/routers/epsilon_greedy/EpsilonGreedy.py b/examples/routers/epsilon_greedy/EpsilonGreedy.py
index b1d25bc893..6cafd763bb 100644
--- a/examples/routers/epsilon_greedy/EpsilonGreedy.py
+++ b/examples/routers/epsilon_greedy/EpsilonGreedy.py
@@ -1,7 +1,7 @@
 import random
 import numpy as np
 
-__version__ = "v1.1"
+__version__ = "v1.2"
 
 def n_success_failures(features,reward):
     n_predictions = features.shape[0]
@@ -10,9 +10,9 @@ def n_success_failures(features,reward):
     return n_success, n_failures
 
 class EpsilonGreedy(object):
-    
+
     def __init__(self,n_branches=None,epsilon=0.1,verbose=False):
-        print "Starting Epsilon Greedy Microservice, version {}".format(__version__)
+        print("Starting Epsilon Greedy Microservice, version {}".format(__version__))
         if n_branches is None:
             raise Exception("n_branches parameter must be given")
         self.verbose = verbose
@@ -22,29 +22,29 @@ def __init__(self,n_branches=None,epsilon=0.1,verbose=False):
         self.branches_tries = [0 for _ in range(n_branches)]
         self.n_branches = n_branches
         if self.verbose:
-            print "Router initialised"
-            print "# branches:",self.n_branches
-            print "Epsilon:",self.epsilon
-            print
-        
+            print("Router initialised")
+            print("# branches:",self.n_branches)
+            print("Epsilon:",self.epsilon)
+            print()
+
     def route(self,features,feature_names):
         x = random.random()
         best_branch = self.best_branch
         other_branches = [i for i in range(self.n_branches) if i!=best_branch]
         selected_branch = best_branch if x>self.epsilon else random.choice(other_branches)
         if self.verbose:
-            print "Routing"
-            print "Current best branch:",best_branch
-            print "Selected branch:",selected_branch
-            print
+            print("Routing")
+            print("Current best branch:",best_branch)
+            print("Selected branch:",selected_branch)
+            print()
         return selected_branch
-        
+
     def send_feedback(self,features,feature_names,routing,reward,truth):
         if self.verbose:
-            print "Training"
-            print "Prev success #", self.branches_success
-            print "Prev tries #", self.branches_tries
-            print "Prev best branch:", self.best_branch
+            print("Training")
+            print("Prev success #", self.branches_success)
+            print("Prev tries #", self.branches_tries)
+            print("Prev best branch:", self.best_branch)
         n_success, n_failures = n_success_failures(features,reward)
         self.branches_success[routing] += n_success
         self.branches_tries[routing] += n_success + n_failures
@@ -55,8 +55,7 @@ def send_feedback(self,features,feature_names,routing,reward,truth):
         ]
         self.best_branch = np.argmax(perfs)
         if self.verbose:
-            print "New success #", self.branches_success
-            print "New tries #", self.branches_tries
-            print "New best branch:",self.best_branch
-            print
-
+            print("New success #", self.branches_success)
+            print("New tries #", self.branches_tries)
+            print("New best branch:",self.best_branch)
+            print()
diff --git a/examples/routers/epsilon_greedy/README.md b/examples/routers/epsilon_greedy/README.md
index 3478efd6fd..549ac73f17 100644
--- a/examples/routers/epsilon_greedy/README.md
+++ b/examples/routers/epsilon_greedy/README.md
@@ -1,10 +1,22 @@
 # Epsilon Greedy Router
 
+## Description
+
+An epsilon-greedy router implements a [multi-armed bandit strategy](https://en.wikipedia.org/wiki/Multi-armed_bandit#Semi-uniform_strategies) in which, when presented with *n* models to make predictions, the currently
+best performing model is selected with probability *1-e* while a random model is selected with probability *e*.
+This strategy ensures sending traffic to the best performing model most of the time (exploitation) while allowing for
+some evaluation of other models (exploration). A typical parameter value could be *e=0.1*, but this will depend on the
+desired trade-off between exploration and exploitation.
+
+Note that in this implementation the parameter value *e* is static, but a related strategy called *epsilon-decreasing*
+would see the value of *e* decrease as the number of predictions increases, resulting in a highly explorative behaviour
+at the start and increasingly exploitative behaviour as time goes on.
+
 
 ## Wrap using s2i
 
 ```bash
-s2i build . seldonio/seldon-core-s2i-python2 egreedy-router
+s2i build . seldonio/seldon-core-s2i-python3 egreedy-router
 ```
 
 ## Smoke Test
@@ -20,4 +32,8 @@ Send a data request.
 ```bash
 data='{"data":{"names":["a","b"],"ndarray":[[1.0,2.0]]}}'
 curl -d "json=${data}" http://0.0.0.0:5000/route
-```
\ No newline at end of file
+```
+
+## Running on Seldon
+An end-to-end example deploying an epsilon-greedy router to route traffic to 3 models in parallel is available [here](
+https://github.com/SeldonIO/seldon-core/blob/master/notebooks/epsilon_greedy_gcp.ipynb).
diff --git a/examples/routers/epsilon_greedy/requirements.txt b/examples/routers/epsilon_greedy/requirements.txt
index 0193419ad2..678d0e5d52 100644
--- a/examples/routers/epsilon_greedy/requirements.txt
+++ b/examples/routers/epsilon_greedy/requirements.txt
@@ -1 +1 @@
-numpy==1.11.2
+numpy==1.14.3
diff --git a/integrations/nvidia-inference-server/.s2i/environment b/integrations/nvidia-inference-server/.s2i/environment
new file mode 100644
index 0000000000..00aba0e6f0
--- /dev/null
+++ b/integrations/nvidia-inference-server/.s2i/environment
@@ -0,0 +1,4 @@
+MODEL_NAME=TRTProxy
+API_TYPE=REST
+SERVICE_TYPE=MODEL
+PERSISTENCE=0
diff --git a/integrations/nvidia-inference-server/Makefile b/integrations/nvidia-inference-server/Makefile
new file mode 100644
index 0000000000..cac653c292
--- /dev/null
+++ b/integrations/nvidia-inference-server/Makefile
@@ -0,0 +1,12 @@
+IMAGE_VERSION=0.1
+IMAGE_NAME = docker.io/seldonio/nvidia-inference-server-proxy
+
+SELDON_CORE_DIR=../../..
+
+.PHONY: build
+build: 
+	s2i build . seldonio/seldon-core-s2i-python3-tensorrt:0.1 $(IMAGE_NAME):$(IMAGE_VERSION)
+
+push_to_dockerhub:
+	docker push $(IMAGE_NAME):$(IMAGE_VERSION)
+
diff --git a/integrations/nvidia-inference-server/README.md b/integrations/nvidia-inference-server/README.md
new file mode 100644
index 0000000000..2ab796bb45
--- /dev/null
+++ b/integrations/nvidia-inference-server/README.md
@@ -0,0 +1,167 @@
+# NVIDIA Inference Server Proxy
+
+The NVIDIA Inference Server Proxy provides a proxy to forward Seldon prediction requests to a running [NVIDIA Inference Server](https://docs.nvidia.com/deeplearning/sdk/inference-user-guide/index.html).
+
+## Configuration
+
+The Nvidia Proxy takes several parameters:
+
+ | Parameter | Type | Value | Example |
+ |-----------|------|-------|---------|
+ | url | STRING | URL to Nvidia Inference Server endpoint | 127.0.0.1:8000 |
+ | model_name | STRING | model name | tensorrt_mnist |
+ | protocol | STRING | API protocol to use: HTTP or GRPC | HTTP |
+
+
+An example SeldonDeployment Kubernetes resource taken from the MNIST demo is shown below to illustrate how these parameters are set. The graph consists of three containers
+
+  1. A Seldon transformer to do feature transformations on the raw input.
+  1. A NVIDIA Inference Server Model Proxy.
+  1. The NVIDIA Inference Server loaded with a model.
+
+![MNIST Example](./mnist-graph.png)
+
+
+```
+{
+    "apiVersion": "machinelearning.seldon.io/v1alpha2",
+    "kind": "SeldonDeployment",
+    "metadata": {
+        "labels": {
+            "app": "seldon"
+        },
+        "name": "nvidia-mnist",
+	"namespace": "kubeflow"
+    },
+    "spec": {
+        "name": "caffe2-mnist",
+        "predictors": [
+            {
+                "componentSpecs": [{
+                    "spec": {
+                        "containers": [
+                            {
+                                "image": "seldonio/mnist-caffe2-transformer:0.1",
+                                "name": "mnist-transformer"
+                            },
+                            {
+                                "image": "seldonio/nvidia-inference-server-proxy:0.1",
+                                "name": "nvidia-proxy"
+                            },
+			    {
+				"args": [
+				    "--model-store=gs://seldon-inference-server-model-store"
+				],
+				"command": [
+				    "inference_server"
+				],
+				"image": "nvcr.io/nvidia/inferenceserver:18.08.1-py2",
+				"livenessProbe": {
+				    "failureThreshold": 3,
+				    "handler":{
+					"httpGet": {
+					    "path": "/api/health/live",
+					    "port": 8000,
+					    "scheme": "HTTP"
+					}
+				    },
+				    "initialDelaySeconds": 5,
+				    "periodSeconds": 5,
+				    "successThreshold": 1,
+				    "timeoutSeconds": 1
+				},
+				"name": "inference-server",
+				"ports": [
+				    {
+					"containerPort": 8000,
+					"protocol": "TCP"
+				    },
+				    {
+					"containerPort": 8001,
+					"protocol": "TCP"
+				    },
+				    {
+					"containerPort": 8002,
+					"protocol": "TCP"
+				    }
+				],
+				"readinessProbe": {
+				    "failureThreshold": 3,
+				    "handler":{
+					"httpGet": {
+					    "path": "/api/health/ready",
+					    "port": 8000,
+					    "scheme": "HTTP"
+					}
+				    },
+				    "initialDelaySeconds": 5,
+				    "periodSeconds": 5,
+				    "successThreshold": 1,
+				    "timeoutSeconds": 1
+				},
+				"resources": {
+				    "limits": {
+					"nvidia.com/gpu": "1"
+				    },
+				    "requests": {
+					"cpu": "100m",
+					"nvidia.com/gpu": "1"
+				    }
+				},
+				"securityContext": {
+				    "runAsUser": 1000
+				}
+			    }
+			],
+			"terminationGracePeriodSeconds": 1,
+			"imagePullSecrets": [
+			    {
+				"name": "ngc"
+			    }
+			]
+		    }
+		}],
+                "graph": {
+                    "name": "mnist-transformer",
+                    "endpoint": { "type" : "REST" },
+                    "type": "TRANSFORMER",
+		    "children": [
+			{
+			    "name": "nvidia-proxy",
+			    "endpoint": { "type" : "REST" },
+			    "type": "MODEL",
+			    "children": [],
+			    "parameters":
+			    [
+				{
+				    "name":"url",
+				    "type":"STRING",
+				    "value":"127.0.0.1:8000"
+				},
+				{
+				    "name":"model_name",
+				    "type":"STRING",
+				    "value":"tensorrt_mnist"
+				},
+				{
+				    "name":"protocol",
+				    "type":"STRING",
+				    "value":"HTTP"
+				}
+			    ]
+			}
+		    ]
+                },
+                "name": "mnist-nvidia",
+                "replicas": 1
+            }
+        ]
+    }
+}
+
+
+```
+
+Examples: 
+
+ * [MNIST with Nvidia Inference Server](../../examples/models/nvidia-mnist/nvidia_mnist.ipynb).
diff --git a/integrations/nvidia-inference-server/TRTProxy.py b/integrations/nvidia-inference-server/TRTProxy.py
new file mode 100644
index 0000000000..3b68c4f70c
--- /dev/null
+++ b/integrations/nvidia-inference-server/TRTProxy.py
@@ -0,0 +1,81 @@
+from inference_server.api import *
+import inference_server.api.model_config_pb2 as model_config
+
+
+def model_dtype_to_np(model_dtype):
+    '''
+    Helper function from https://github.com/NVIDIA/dl-inference-server/blob/18.08/src/clients/python/image_client.py
+    '''
+    if model_dtype == model_config.TYPE_BOOL:
+        return np.bool
+    elif model_dtype == model_config.TYPE_INT8:
+        return np.int8
+    elif model_dtype == model_config.TYPE_INT16:
+        return np.int16
+    elif model_dtype == model_config.TYPE_INT32:
+        return np.int32
+    elif model_dtype == model_config.TYPE_INT64:
+        return np.int64
+    elif model_dtype == model_config.TYPE_UINT8:
+        return np.uint8
+    elif model_dtype == model_config.TYPE_UINT16:
+        return np.uint16
+    elif model_dtype == model_config.TYPE_FP16:
+        return np.float16
+    elif model_dtype == model_config.TYPE_FP32:
+        return np.float32
+    elif model_dtype == model_config.TYPE_FP64:
+        return np.float64
+    return None
+
+def parse_model(url, protocol, model_name, verbose=False):
+    ctx = ServerStatusContext(url, protocol, model_name, verbose)
+    server_status = ctx.get_server_status()
+
+    if model_name not in server_status.model_status:
+        raise Exception("unable to find model:"+model_name)
+
+    status = server_status.model_status[model_name]
+    config = status.config
+
+    input = config.input[0]
+    output = config.output[0]
+
+    return (input.name, output.name, model_dtype_to_np(input.data_type), input.dims)
+
+
+'''
+A basic tensorflow serving proxy
+'''
+class TRTProxy(object):
+
+    def __init__(self,url=None,protocol="HTTP",model_name=None,model_version=1):
+        print("URL:",url)
+        self.url = url
+        self.protocol_id = ProtocolType.from_str(protocol)
+        self.model_version = model_version
+        if protocol == "GRPC":
+            self.grpc = True
+            channel = grpc.insecure_channel(url)
+            #self.stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
+        else:
+            self.grpc = False
+        self.model_name = model_name
+        self.input_name, self.output_name, self.dtype, self.input_dims = parse_model(url, self.protocol_id, model_name, False)
+        self.ctx = InferContext(self.url, self.protocol_id,self.model_name, self.model_version, False)
+
+        
+    
+    def predict(self,X,features_names):
+        X = X.astype(self.dtype)
+        if self.grpc:
+            print("not implemented")
+        else:
+            if len(X.shape) == len(self.input_dims):
+                X = [X]
+            results = self.ctx.run(
+                { self.input_name : X },
+                { self.output_name : InferContext.ResultFormat.RAW },
+                1)
+            return results[self.output_name]
+        return []
diff --git a/integrations/nvidia-inference-server/mnist-graph.png b/integrations/nvidia-inference-server/mnist-graph.png
new file mode 100644
index 0000000000..d78dab25fe
Binary files /dev/null and b/integrations/nvidia-inference-server/mnist-graph.png differ
diff --git a/integrations/tfserving/.s2i/environment b/integrations/tfserving/.s2i/environment
new file mode 100644
index 0000000000..faf6e43bf8
--- /dev/null
+++ b/integrations/tfserving/.s2i/environment
@@ -0,0 +1,4 @@
+MODEL_NAME=TfServingProxy
+API_TYPE=REST
+SERVICE_TYPE=MODEL
+PERSISTENCE=0
diff --git a/integrations/tfserving/Makefile b/integrations/tfserving/Makefile
new file mode 100644
index 0000000000..3565a1b4f6
--- /dev/null
+++ b/integrations/tfserving/Makefile
@@ -0,0 +1,12 @@
+IMAGE_VERSION=0.1
+IMAGE_NAME = docker.io/seldonio/tfserving-proxy
+
+SELDON_CORE_DIR=../../..
+
+.PHONY: build
+build: 
+	s2i build . seldonio/seldon-core-s2i-python3:0.2 $(IMAGE_NAME):$(IMAGE_VERSION)
+
+push_to_dockerhub:
+	docker push $(IMAGE_NAME):$(IMAGE_VERSION)
+
diff --git a/integrations/tfserving/README.md b/integrations/tfserving/README.md
new file mode 100644
index 0000000000..11a82395a2
--- /dev/null
+++ b/integrations/tfserving/README.md
@@ -0,0 +1,122 @@
+# TensorFlow Serving Proxy
+
+The TensorFlow Serving Proxy provides a proxy to forward Seldon prediction requests to a running [TenserFlow Serving](https://www.tensorflow.org/serving/) server.
+
+## Configuration
+
+The tensorflow proxy takes several parameters:
+
+ | Parameter | Type | Value | Example |
+ |-----------|------|-------|---------|
+ | rest_endpoint | STRING | URL of server HTTP endpoint | http://0.0.0.0:8000 |
+ | grpc_endpoint | STRING | host and port of gRPC endpoint | 0.0.0.0:8001 |
+ | model_name | STRING | model name | mnist-model |
+ | signature_name | STRING | model signature name | predict_images |
+ | model_input | STRING | model input name | images |
+ | model_output | STRING | model output name | scores |
+
+An example resource with the proxy and a Tensorflow Serving server is shown below.
+
+
+```
+{
+    "apiVersion": "machinelearning.seldon.io/v1alpha2",
+    "kind": "SeldonDeployment",
+    "metadata": {
+        "labels": {
+            "app": "seldon"
+        },
+        "name": "tfserving-mnist",
+	"namespace": "seldon"	
+    },
+    "spec": {
+        "name": "tf-mnist",
+        "predictors": [
+            {
+                "componentSpecs": [{
+                    "spec": {
+                        "containers": [
+                            {
+                                "image": "seldonio/tfserving-proxy:0.1",
+                                "name": "tfserving-proxy"
+                            },
+			    {
+				"args": [
+				    "/usr/bin/tensorflow_model_server",
+				    "--port=8000",
+				    "--model_name=mnist-model",
+				    "--model_base_path=gs://seldon-tfserving-store/mnist-model"
+				],
+				"image": "gcr.io/kubeflow-images-public/tensorflow-serving-1.7:v20180604-0da89b8a",
+				"name": "mnist-model",
+				"ports": [
+				    {
+					"containerPort": 8000,
+					"protocol": "TCP"
+				    }
+				],
+				"resources": {
+				    "limits": {
+					"cpu": "4",
+					"memory": "4Gi"
+				    },
+				    "requests": {
+					"cpu": "1",
+					"memory": "1Gi"
+				    }
+				},
+				"securityContext": {
+				    "runAsUser": 1000
+				}
+			    }
+			],
+			"terminationGracePeriodSeconds": 1
+		    }
+		}],
+                "graph": {
+		    "name": "tfserving-proxy",
+		    "endpoint": { "type" : "REST" },
+		    "type": "MODEL",
+		    "children": [],
+		    "parameters":
+		    [
+			{
+			    "name":"grpc_endpoint",
+			    "type":"STRING",
+			    "value":"localhost:8000"
+			},
+			{
+			    "name":"model_name",
+			    "type":"STRING",
+			    "value":"mnist-model"
+			},
+			{
+			    "name":"model_output",
+			    "type":"STRING",
+			    "value":"scores"
+			},
+			{
+			    "name":"model_input",
+			    "type":"STRING",
+			    "value":"images"
+			},
+			{
+			    "name":"signature_name",
+			    "type":"STRING",
+			    "value":"predict_images"
+			}
+		    ]
+		},
+                "name": "mnist-tfserving",
+                "replicas": 1
+            }
+        ]
+    }
+}
+
+
+```
+
+Examples:
+
+ * [MNIST with TensorFlow Serving Proxy](../../examples/models/tfserving-mnist/tfserving-mnist.ipynb).
diff --git a/integrations/tfserving/TfServingProxy.py b/integrations/tfserving/TfServingProxy.py
new file mode 100644
index 0000000000..15e88a252f
--- /dev/null
+++ b/integrations/tfserving/TfServingProxy.py
@@ -0,0 +1,69 @@
+import grpc
+import numpy
+import tensorflow as tf
+
+from tensorflow.python.saved_model import signature_constants
+from tensorflow_serving.apis import predict_pb2
+from tensorflow_serving.apis import prediction_service_pb2_grpc
+
+import requests
+import json
+
+class TensorflowServerError(Exception):
+
+    def __init__(self, message):
+        self.message = message
+
+'''
+A basic tensorflow serving proxy
+'''
+class TfServingProxy(object):
+
+    def __init__(self,rest_endpoint=None,grpc_endpoint=None,model_name=None,signature_name=None,model_input=None,model_output=None):
+        print("rest_endpoint:",rest_endpoint)
+        print("grpc_endpoint:",grpc_endpoint)
+        if not grpc_endpoint is None:
+            self.grpc = True
+            channel = grpc.insecure_channel(grpc_endpoint)
+            self.stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
+        else:
+            self.grpc = False
+            self.rest_endpoint = rest_endpoint
+        self.model_name = model_name
+        if signature_name is None:
+            self.signature_name = signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY
+        else:
+            self.signature_name = signature_name
+        self.model_input = model_input        
+        self.model_output = model_output
+        
+    def predict(self,X,features_names):
+        if self.grpc:
+            request = predict_pb2.PredictRequest()
+            request.model_spec.name = self.model_name
+            request.model_spec.signature_name = self.signature_name
+            request.inputs[self.model_input].CopyFrom(tf.contrib.util.make_tensor_proto(X.tolist(), shape=X.shape))
+            print(request)
+            result = self.stub.Predict(request)
+            print(result)
+            response = numpy.array(result.outputs[self.model_output].float_val)
+            if len(response.shape) == 1:
+                response = numpy.expand_dims(response, axis=0)
+            return response
+        else:
+            print(self.rest_endpoint)
+            data = {"instances":X.tolist()}
+            if not self.signature_name is None:
+                data["signature_name"] = self.signature_name
+            print(data)
+            response = requests.post(
+                self.rest_endpoint,
+                data = json.dumps(data))
+            if response.status_code == 200:
+                result = numpy.array(response.json()["predictions"])
+                if len(result.shape) == 1:
+                    result = numpy.expand_dims(result, axis=0)
+                return result
+            else:
+                print("Error from server:",response)
+                raise TensorflowServerError(response.json())
diff --git a/integrations/tfserving/requirements.txt b/integrations/tfserving/requirements.txt
new file mode 100644
index 0000000000..78b3517253
--- /dev/null
+++ b/integrations/tfserving/requirements.txt
@@ -0,0 +1,3 @@
+tensorflow==1.10.1
+tensorflow-serving-api==1.10.1
+requests
diff --git a/notebooks/kubectl_demo_minikube_rbac.ipynb b/notebooks/kubectl_demo_minikube_rbac.ipynb
index 84fef8574e..a2381ff3fc 100644
--- a/notebooks/kubectl_demo_minikube_rbac.ipynb
+++ b/notebooks/kubectl_demo_minikube_rbac.ipynb
@@ -34,6 +34,12 @@
     "Your start command would then look like:\n",
     "```\n",
     "minikube start --vm-driver kvm2 --memory 4096 --feature-gates=CustomResourceValidation=true --extra-config=apiserver.Authorization.Mode=RBAC\n",
+    "```\n",
+    "**2018-09-24** : Alternatively, you can try this command with a new formatting of --extra-config parameter otherwise your start command will hang forever (see https://github.com/kubernetes/minikube/issues/2798#issuecomment-420402313 )\n",
+    "\n",
+    "Your start command would then look like:\n",
+    "```\n",
+    "minikube start --vm-driver kvm2 --memory 4096 --feature-gates=CustomResourceValidation=true --extra-config=apiserver.authorization-mode=RBAC\n",
     "```"
    ]
   },
diff --git a/notebooks/visualizer.py b/notebooks/visualizer.py
index 405bbb2750..9792a98ab1 100644
--- a/notebooks/visualizer.py
+++ b/notebooks/visualizer.py
@@ -21,16 +21,11 @@ def get_graph(filename,predictor=0):
     deployment = json.load(open(filename,'r'))
     predictors = deployment.get("spec").get("predictors")
     dot = graphviz.Digraph()
-    
-    with dot.subgraph(name="cluster_0") as pdot:
-        graph = predictors[0].get("graph")
-        _populate_graph(pdot, graph, suffix='0')
-        pdot.attr(label="predictor")
-        
-    if len(predictors)>1:
-        with dot.subgraph(name="cluster_1") as cdot:
-            graph = predictors[1].get("graph")
-            _populate_graph(cdot, graph, suffix='1')
-            cdot.attr(label="canary")
-        
+
+    for idx in range(len(predictors)):
+        with dot.subgraph(name="cluster_"+str(idx)) as pdot:
+            graph = predictors[idx].get("graph")
+            _populate_graph(pdot, graph, suffix=str(idx))
+            pdot.attr(label="predictor-"+str(idx))
     return dot
+
diff --git a/readme.md b/readme.md
index 386c5c9cdc..e0d06c4410 100644
--- a/readme.md
+++ b/readme.md
@@ -114,7 +114,8 @@ Seldon-core allows various types of components to be built and plugged into the
        * [Train and deploy a PyTorch MNIST classififer using FfDL and Seldon.](https://github.com/IBM/FfDL/blob/master/community/FfDL-Seldon/pytorch-model/README.md)
  * [Istio and Seldon](./docs/istio.md)
    * [Canary deployemts using Istio and Seldon.](examples/istio/canary_update/canary.ipynb).
-	
+ * [NVIDIA TensorRT and DL Inference Server](./integrations/nvidia-inference-server/README.md)
+ * [Tensorflow Serving](./integrations/tfserving/README.md)	
 
 ## Install
 
@@ -190,7 +191,8 @@ Three steps:
 | [Seldon Java Runtime Wrapper for S2I](docs/wrappers/java.md) | [seldonio/seldon-core-s2i-java-runtime](https://hub.docker.com/r/seldonio/seldon-core-s2i-java-runtime/tags/) | 0.1 | | 
 | [Seldon R Wrapper for S2I](docs/wrappers/r.md) | [seldonio/seldon-core-s2i-r](https://hub.docker.com/r/seldonio/seldon-core-s2i-r/tags/) | 0.1 | |
 | [Seldon NodeJS Wrapper for S2I](docs/wrappers/nodejs.md) | [seldonio/seldon-core-s2i-nodejs](https://hub.docker.com/r/seldonio/seldon-core-s2i-nodejs/tags/) | 0.1 | |
-
+| [Seldon Tensorflow Serving proxy](integrations/tfserving/README.md) | [seldonio/tfserving-proxy](https://hub.docker.com/r/seldonio/tfserving-proxy/tags/) | 0.1 | 
+| [Seldon NVIDIA inference server proxy](integrations/nvidia-inference-server/README.md) | [seldonio/nvidia-inference-server-proxy](https://hub.docker.com/r/seldonio/nvidia-inference-server-proxy/tags/) | 0.1 | 
 #### Java Packages
 
 | Description | Package | Version |
diff --git a/seldon-core/seldon-core/core.libsonnet b/seldon-core/seldon-core/core.libsonnet
index 767c574d98..552ff6bc87 100644
--- a/seldon-core/seldon-core/core.libsonnet
+++ b/seldon-core/seldon-core/core.libsonnet
@@ -96,6 +96,7 @@ local getEnvNotRedis(x) = x.name != "SELDON_CLUSTER_MANAGER_REDIS_HOST";
         { name: "JAVA_OPTS", value: javaOpts },
         { name: "SPRING_OPTS", value: springOpts },
         { name: "ENGINE_CONTAINER_IMAGE_AND_VERSION", value: engineImage },
+        { name: "ENGINE_CONTAINER_IMAGE_PULL_POLICY", value: "IfNotPresent" },
         { name: "SELDON_CLUSTER_MANAGER_REDIS_HOST", value: name+"-redis" },
         { name: "SELDON_CLUSTER_MANAGER_POD_NAMESPACE", valueFrom: {fieldRef:{apiVersion: "v1",fieldPath: "metadata.namespace"}}},
       ];
diff --git a/seldon-core/seldon-core/json/template.json b/seldon-core/seldon-core/json/template.json
index 38e3614b2d..7f4f67b4fc 100644
--- a/seldon-core/seldon-core/json/template.json
+++ b/seldon-core/seldon-core/json/template.json
@@ -8,16 +8,16 @@
             "apiVersion": "v1",
             "metadata": {
                 "name": "seldon",
-                "namespace": "seldon",
+                "namespace": "default",
                 "creationTimestamp": null
             }
         },
         {
             "kind": "Role",
-            "apiVersion": "rbac.authorization.k8s.io/v1beta1",
+            "apiVersion": "rbac.authorization.k8s.io/v1",
             "metadata": {
                 "name": "seldon-local",
-                "namespace": "seldon",
+                "namespace": "default",
                 "creationTimestamp": null
             },
             "rules": [
@@ -48,7 +48,7 @@
         },
         {
             "kind": "ClusterRole",
-            "apiVersion": "rbac.authorization.k8s.io/v1beta1",
+            "apiVersion": "rbac.authorization.k8s.io/v1",
             "metadata": {
                 "name": "seldon-crd",
                 "creationTimestamp": null
@@ -72,14 +72,14 @@
             "apiVersion": "rbac.authorization.k8s.io/v1",
             "metadata": {
                 "name": "seldon",
-                "namespace": "seldon",
+                "namespace": "default",
                 "creationTimestamp": null
             },
             "subjects": [
                 {
                     "kind": "ServiceAccount",
                     "name": "seldon",
-                    "namespace": "seldon"
+                    "namespace": "default"
                 }
             ],
             "roleRef": {
@@ -99,7 +99,7 @@
                 {
                     "kind": "ServiceAccount",
                     "name": "seldon",
-                    "namespace": "seldon"
+                    "namespace": "default"
                 }
             ],
             "roleRef": {
@@ -110,7 +110,7 @@
         },
         {
             "kind": "Role",
-            "apiVersion": "rbac.authorization.k8s.io/v1beta1",
+            "apiVersion": "rbac.authorization.k8s.io/v1",
             "metadata": {
                 "name": "ambassador",
                 "creationTimestamp": null
@@ -162,7 +162,7 @@
         },
         {
             "kind": "RoleBinding",
-            "apiVersion": "rbac.authorization.k8s.io/v1beta1",
+            "apiVersion": "rbac.authorization.k8s.io/v1",
             "metadata": {
                 "name": "ambassador",
                 "creationTimestamp": null
@@ -171,7 +171,7 @@
                 {
                     "kind": "ServiceAccount",
                     "name": "seldon",
-                    "namespace": "seldon"
+                    "namespace": "default"
                 }
             ],
             "roleRef": {
@@ -245,7 +245,7 @@
         },
         {
             "kind": "Deployment",
-            "apiVersion": "extensions/v1beta1",
+            "apiVersion": "apps/v1",
             "metadata": {
                 "name": "RELEASE-NAME-ambassador",
                 "creationTimestamp": null,
@@ -384,16 +384,16 @@
         },
         {
             "kind": "Deployment",
-            "apiVersion": "apps/v1beta1",
+            "apiVersion": "apps/v1",
             "metadata": {
                 "name": "RELEASE-NAME-seldon-apiserver",
-                "namespace": "seldon",
+                "namespace": "default",
                 "creationTimestamp": null,
                 "labels": {
                     "app": "seldon-apiserver-container-app",
                     "app.kubernetes.io/component": "seldon-core-apiserver",
                     "app.kubernetes.io/name": "RELEASE-NAME",
-                    "chart": "seldon-core-0.2.3-SNAPSHOT",
+                    "chart": "seldon-core-0.2.4-SNAPSHOT",
                     "component": "seldon-core",
                     "heritage": "Tiller",
                     "release": "RELEASE-NAME"
@@ -406,7 +406,7 @@
                         "app": "seldon-apiserver-container-app",
                         "app.kubernetes.io/component": "seldon-core-apiserver",
                         "app.kubernetes.io/name": "RELEASE-NAME",
-                        "chart": "seldon-core-0.2.3-SNAPSHOT",
+                        "chart": "seldon-core-0.2.4-SNAPSHOT",
                         "component": "seldon-core",
                         "heritage": "Tiller",
                         "release": "RELEASE-NAME"
@@ -419,7 +419,7 @@
                             "app": "seldon-apiserver-container-app",
                             "app.kubernetes.io/component": "seldon-core-apiserver",
                             "app.kubernetes.io/name": "RELEASE-NAME",
-                            "chart": "seldon-core-0.2.3-SNAPSHOT",
+                            "chart": "seldon-core-0.2.4-SNAPSHOT",
                             "component": "seldon-core",
                             "heritage": "Tiller",
                             "release": "RELEASE-NAME"
@@ -451,7 +451,7 @@
                         "containers": [
                             {
                                 "name": "seldon-apiserver-container",
-                                "image": "seldonio/apife:0.2.3-SNAPSHOT",
+                                "image": "seldonio/apife:0.2.4-SNAPSHOT",
                                 "ports": [
                                     {
                                         "containerPort": 8080,
@@ -557,16 +557,16 @@
         },
         {
             "kind": "Deployment",
-            "apiVersion": "extensions/v1beta1",
+            "apiVersion": "apps/v1",
             "metadata": {
                 "name": "RELEASE-NAME-seldon-cluster-manager",
-                "namespace": "seldon",
+                "namespace": "default",
                 "creationTimestamp": null,
                 "labels": {
                     "app": "seldon-cluster-manager-server",
                     "app.kubernetes.io/component": "seldon-core-operator",
                     "app.kubernetes.io/name": "RELEASE-NAME",
-                    "chart": "seldon-core-0.2.3-SNAPSHOT",
+                    "chart": "seldon-core-0.2.4-SNAPSHOT",
                     "component": "seldon-core",
                     "heritage": "Tiller",
                     "release": "RELEASE-NAME"
@@ -579,7 +579,7 @@
                         "app": "seldon-cluster-manager-server",
                         "app.kubernetes.io/component": "seldon-core-operator",
                         "app.kubernetes.io/name": "RELEASE-NAME",
-                        "chart": "seldon-core-0.2.3-SNAPSHOT",
+                        "chart": "seldon-core-0.2.4-SNAPSHOT",
                         "component": "seldon-core",
                         "heritage": "Tiller",
                         "release": "RELEASE-NAME"
@@ -592,7 +592,7 @@
                             "app": "seldon-cluster-manager-server",
                             "app.kubernetes.io/component": "seldon-core-operator",
                             "app.kubernetes.io/name": "RELEASE-NAME",
-                            "chart": "seldon-core-0.2.3-SNAPSHOT",
+                            "chart": "seldon-core-0.2.4-SNAPSHOT",
                             "component": "seldon-core",
                             "heritage": "Tiller",
                             "release": "RELEASE-NAME"
@@ -602,7 +602,7 @@
                         "containers": [
                             {
                                 "name": "seldon-cluster-manager-container",
-                                "image": "seldonio/cluster-manager:0.2.3-SNAPSHOT",
+                                "image": "seldonio/cluster-manager:0.2.4-SNAPSHOT",
                                 "ports": [
                                     {
                                         "containerPort": 8080,
@@ -622,7 +622,11 @@
                                     },
                                     {
                                         "name": "ENGINE_CONTAINER_IMAGE_AND_VERSION",
-                                        "value": "seldonio/engine:0.2.3-SNAPSHOT"
+                                        "value": "seldonio/engine:0.2.4-SNAPSHOT"
+                                    },
+                                    {
+                                        "name": "ENGINE_CONTAINER_IMAGE_PULL_POLICY",
+                                        "value": "IfNotPresent"
                                     },
                                     {
                                         "name": "SELDON_CLUSTER_MANAGER_POD_NAMESPACE",
@@ -644,7 +648,7 @@
                             }
                         ],
                         "restartPolicy": "Always",
-                        "terminationGracePeriodSeconds": 30,
+                        "terminationGracePeriodSeconds": 1,
                         "dnsPolicy": "ClusterFirst",
                         "serviceAccountName": "seldon",
                         "serviceAccount": "seldon",
@@ -664,16 +668,16 @@
         },
         {
             "kind": "Deployment",
-            "apiVersion": "extensions/v1beta1",
+            "apiVersion": "apps/v1",
             "metadata": {
                 "name": "RELEASE-NAME-redis",
-                "namespace": "seldon",
+                "namespace": "default",
                 "creationTimestamp": null,
                 "labels": {
                     "app": "RELEASE-NAME-redis-app",
                     "app.kubernetes.io/component": "seldon-core-redis",
                     "app.kubernetes.io/name": "RELEASE-NAME",
-                    "chart": "seldon-core-0.2.3-SNAPSHOT",
+                    "chart": "seldon-core-0.2.4-SNAPSHOT",
                     "component": "seldon-core",
                     "heritage": "Tiller",
                     "release": "RELEASE-NAME"
@@ -686,7 +690,7 @@
                         "app": "RELEASE-NAME-redis-app",
                         "app.kubernetes.io/component": "seldon-core-redis",
                         "app.kubernetes.io/name": "RELEASE-NAME",
-                        "chart": "seldon-core-0.2.3-SNAPSHOT",
+                        "chart": "seldon-core-0.2.4-SNAPSHOT",
                         "component": "seldon-core",
                         "heritage": "Tiller",
                         "release": "RELEASE-NAME"
@@ -699,7 +703,7 @@
                             "app": "RELEASE-NAME-redis-app",
                             "app.kubernetes.io/component": "seldon-core-redis",
                             "app.kubernetes.io/name": "RELEASE-NAME",
-                            "chart": "seldon-core-0.2.3-SNAPSHOT",
+                            "chart": "seldon-core-0.2.4-SNAPSHOT",
                             "component": "seldon-core",
                             "heritage": "Tiller",
                             "release": "RELEASE-NAME"
diff --git a/wrappers/s2i/nodejs/Dockerfile b/wrappers/s2i/nodejs/Dockerfile
index 440f304593..7f895e4a8f 100644
--- a/wrappers/s2i/nodejs/Dockerfile
+++ b/wrappers/s2i/nodejs/Dockerfile
@@ -5,9 +5,13 @@ LABEL io.openshift.s2i.scripts-url="image:///s2i/bin"
 RUN mkdir microservice
 WORKDIR /microservice
 
-COPY microservice.js /microservice
+COPY *microservice.js /microservice/
 
-COPY package.json /microservice
+COPY package.json /microservice/
+
+COPY prediction_grpc_pb.js /microservice/
+
+COPY prediction_pb.js /microservice/
 
 RUN npm install
 
diff --git a/wrappers/s2i/nodejs/Makefile b/wrappers/s2i/nodejs/Makefile
index 618cd4f9ed..32cbcfbb91 100644
--- a/wrappers/s2i/nodejs/Makefile
+++ b/wrappers/s2i/nodejs/Makefile
@@ -1,4 +1,4 @@
-VERSION=0.1
+VERSION=0.2-SNAPSHOT
 IMAGE_NAME = docker.io/seldonio/seldon-core-s2i-nodejs:${VERSION}
 
 SELDON_CORE_DIR=../../..
diff --git a/wrappers/s2i/nodejs/README.md b/wrappers/s2i/nodejs/README.md
index ac7c19dc85..97d2a10f58 100644
--- a/wrappers/s2i/nodejs/README.md
+++ b/wrappers/s2i/nodejs/README.md
@@ -57,7 +57,7 @@ This builds the base wrapper image needed for any nodejs model to be deployed on
 ### Building s2i nodejs model Image
 
 ```
-s2i build test/model-template-app seldonio/seldon-core-s2i-nodejs seldon-core-template-model
+s2i build -E ./test/model-template-app/.s2i/environment test/model-template-app seldonio/seldon-core-s2i-nodejs seldon-core-template-model
 ```
 
 This creates the actual nodejs model image as a seldon component
@@ -77,3 +77,22 @@ Make sure the current user can run npm commands.
 ```
 make test
 ```
+
+### GRPC code-generated Proto JS Files
+
+This is the code pre-generated using protoc and the Node gRPC protoc plugin, and the generated code can be found in various `*_pb.js` files.
+The creation of the grpc srevice assumes these files to be present.
+
+```
+cd ../../../proto/
+npm install -g grpc-tools
+grpc_tools_node_protoc --js_out=import_style=commonjs,binary:../wrappers/s2i/nodejs/ --grpc_out=../wrappers/s2i/nodejs --plugin=protoc-gen-grpc=`which grpc_tools_node_protoc_plugin` prediction.proto
+cd ../wrappers/s2i/nodejs/
+```
+
+### Test using GRPC client
+
+```
+npm i
+node grpc_client.js
+```
diff --git a/wrappers/s2i/nodejs/grpc_client.js b/wrappers/s2i/nodejs/grpc_client.js
new file mode 100644
index 0000000000..757b751afe
--- /dev/null
+++ b/wrappers/s2i/nodejs/grpc_client.js
@@ -0,0 +1,61 @@
+/*
+ *
+ * Copyright 2015 gRPC authors.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ *
+ */
+
+var messages = require("./prediction_pb");
+var services = require("./prediction_grpc_pb");
+
+var grpc = require("grpc");
+
+function main() {
+  var client = new services.ModelClient(
+    "localhost:5000",
+    grpc.credentials.createInsecure()
+  );
+  var tensorData = new messages.Tensor();
+  tensorData.setShapeList([1, 10]);
+  tensorData.setValuesList([0, 0, 1, 1, 5, 6, 7, 8, 4, 3]);
+
+  var defdata = new messages.DefaultData();
+  defdata.setTensor(tensorData);
+  defdata.setNamesList([]);
+
+  var request = new messages.SeldonMessage();
+  request.setData(defdata);
+  client.predict(request, function(err, response) {
+    if (err) {
+      console.log(err);
+    } else {
+      console.log(
+        "Seldon Message => \n\nNames: ",
+        response.getData().getNamesList(),
+        "\n\nShape: ",
+        response
+          .getData()
+          .getTensor()
+          .getShapeList(),
+        "\n\nValues: ",
+        response
+          .getData()
+          .getTensor()
+          .getValuesList()
+      );
+    }
+  });
+}
+
+main();
diff --git a/wrappers/s2i/nodejs/microservice.js b/wrappers/s2i/nodejs/microservice.js
index 64d1237120..72b668f352 100644
--- a/wrappers/s2i/nodejs/microservice.js
+++ b/wrappers/s2i/nodejs/microservice.js
@@ -1,23 +1,9 @@
 const argparse = require("argparse");
-const express = require("express");
-const app = express();
-const bodyParser = require("body-parser");
 const nj = require("numjs");
-app.use(bodyParser.urlencoded({ extended: true }));
-let predict = null;
+const grpc_messages = require("./prediction_pb");
+const process = require("process");
 const port = process.env.PREDICTIVE_UNIT_SERVICE_PORT || 5000;
-const loadModel = async function(model) {
-  model = "./model/" + model;
-  try {
-    const MyModel = require(model);
-    console.log("Loading Model", model);
-    let x = new MyModel();
-    await x.init();
-    return x.predict.bind(x);
-  } catch (msg) {
-    return msg;
-  }
-};
+let user_model = null;
 
 const get_predict_classNames = function(size) {
   let className = [];
@@ -30,6 +16,7 @@ const get_predict_classNames = function(size) {
 };
 
 const rest_data_to_array = function(data) {
+  let features = null;
   if (data["tensor"]) {
     features = nj
       .array(data["tensor"]["values"])
@@ -58,6 +45,33 @@ const array_to_rest_data = function(array, original_datadef) {
   return data;
 };
 
+const array_to_grpc_data = function(array, original_datadef) {
+  array = nj.array(array);
+
+  var defdata = new grpc_messages.DefaultData();
+  defdata.setNamesList(get_predict_classNames(array.shape[1]));
+  if (original_datadef["tensor"]) {
+    var tensorData = new grpc_messages.Tensor();
+    tensorData.setShapeList(array.shape.length > 1 ? array.shape : []);
+    tensorData.setValuesList(array.flatten().tolist());
+    defdata.setTensor(tensorData);
+  } else if (original_datadef["ndarray"]) {
+    datadef.setNdarray(array.tolist());
+  } else {
+    datadef.setNdarray(array.tolist());
+  }
+  var data = new grpc_messages.SeldonMessage();
+  data.setData(defdata);
+  return data;
+};
+
+const dataFunctions = [
+  rest_data_to_array,
+  array_to_rest_data,
+  array_to_grpc_data,
+  get_predict_classNames
+];
+
 const parser = new argparse.ArgumentParser({
   description: "Seldon-core nodejs microservice builder",
   addHelp: true
@@ -81,34 +95,53 @@ parser.addArgument("--persistence", {
 });
 const args = parser.parseArgs();
 
-console.log(args.model, args.api, args.service, args.persistence);
+const loadModel = async function(model) {
+  model = "./model/" + model;
+  try {
+    const MyModel = require(model);
+    console.log("Loading Model", model);
+    let x = new MyModel();
+    await x.init();
+    return x;
+  } catch (msg) {
+    return msg;
+  }
+};
 
-if (args.service === "MODEL" && args.api === "REST") {
-  app.post("/predict", async (req, res) => {
-    try {
-      body = JSON.parse(req.body.json);
-      body = body.data;
-    } catch (msg) {
-      console.log(msg);
-      res.status(500).send("Cannot parse predict input json " + req.body);
-    }
-    if (predict && typeof predict === "function") {
-      result = predict(rest_data_to_array(body), body.names);
-      result = { data: array_to_rest_data(result, body) };
-      res.status(200).send(result);
-    } else {
-      console.log("Predict function not Found");
-      res.status(500).send(predict);
-    }
-  });
-}
+const createServer = () => {
+  if (args.service === "MODEL") {
+    require("./model_microservice.js")(
+      user_model,
+      args.api,
+      port,
+      ...dataFunctions
+    );
+  }
+  if (args.service === "TRANSFORMER") {
+    require("./transformer_microservice.js")(
+      user_model,
+      args.api,
+      port,
+      ...dataFunctions
+    );
+  }
+};
 
-app.listen(port, async () => {
-  predict = await loadModel(
+const getModelFunction = async () => {
+  user_model = await loadModel(
     args.model,
     args.api,
     args.service,
     args.persistence
   );
-  console.log(`NodeJs Microservice listening on port ${port}!`);
-});
+  if (user_model) {
+    console.log("Model Class loaded successfully");
+    createServer();
+  } else {
+    console.log("Model Class could not be loaded ", user_model);
+    process.exit(1);
+  }
+};
+
+getModelFunction();
+console.log(args.model, args.api, args.service, args.persistence);
diff --git a/wrappers/s2i/nodejs/model_microservice.js b/wrappers/s2i/nodejs/model_microservice.js
new file mode 100644
index 0000000000..b4409c998c
--- /dev/null
+++ b/wrappers/s2i/nodejs/model_microservice.js
@@ -0,0 +1,155 @@
+const express = require("express");
+const app = express();
+const bodyParser = require("body-parser");
+const grpc = require("grpc");
+const grpc_services = require("./prediction_grpc_pb");
+
+module.exports = (
+  user_model,
+  api,
+  port,
+  rest_data_to_array,
+  array_to_rest_data,
+  array_to_grpc_data
+) => {
+  if (user_model.predict && typeof user_model.predict === "function") {
+    console.log("Predict function loaded successfully");
+  } else if (
+    user_model.send_feedback &&
+    typeof user_model.send_feedback === "function"
+  ) {
+    console.log("Send feedback function loaded successfully");
+  } else {
+    console.log(user_model);
+    process.exit(1);
+  }
+  let predict = user_model.predict ? user_model.predict.bind(user_model) : null;
+  let send_feedback = user_model.send_feedback
+    ? user_model.send_feedback.bind(user_model)
+    : null;
+
+  if (api === "REST") {
+    app.use(bodyParser.urlencoded({ extended: true }));
+    app.post("/predict", (req, res) => {
+      try {
+        body = JSON.parse(req.body.json);
+        body = body.data;
+      } catch (msg) {
+        console.log(msg);
+        res.status(500).send("Cannot parse predict input json " + req.body);
+      }
+      if (predict && typeof predict === "function") {
+        result = predict(rest_data_to_array(body), body.names);
+        result = { data: array_to_rest_data(result, body) };
+        res.status(200).send(result);
+      } else {
+        console.log("Predict function not Found");
+        res.status(500).send(null);
+      }
+    });
+    app.post("/send-feedback", (req, res) => {
+      try {
+        body = JSON.parse(req.body.json);
+        request = body.request;
+        body = request.data;
+      } catch (msg) {
+        console.log(msg);
+        res
+          .status(500)
+          .send("Cannot parse Send feedback input json " + req.body);
+      }
+      if (send_feedback && typeof send_feedback === "function") {
+        result = send_feedback(
+          rest_data_to_array(body),
+          body.names,
+          rest_data_to_array(request.truth),
+          request.reward
+        );
+        // result = { data: array_to_rest_data(result, body) };
+        res.status(200).send({});
+      } else {
+        console.log("Send feedback function not Found");
+        res.status(500).send(null);
+      }
+    });
+    var server = app.listen(port, () => {
+      console.log(`NodeJs REST Microservice listening on port ${port}!`);
+    });
+    function stopServer(code) {
+      server.close();
+      console.log(`About to exit with code: ${code}`);
+    }
+    process.on("SIGINT", stopServer.bind(this));
+    process.on("SIGTERM", stopServer.bind(this));
+  }
+
+  if (api === "GRPC") {
+    function predictEndpoint(call, callback) {
+      let data = call.request.getData();
+      let body = { names: data.getNamesList() };
+
+      if (data.hasTensor()) {
+        data = data.getTensor();
+        body["tensor"] = {
+          shape: data.getShapeList(),
+          values: data.getValuesList()
+        };
+      } else {
+        body["ndarray"] = data.getNdarray();
+      }
+      result = predict(rest_data_to_array(body), body.names);
+      callback(null, array_to_grpc_data(result, body));
+    }
+    function feedbackEndpoint(call, callback) {
+      let request = call.request.getRequest();
+      let data = request.getData();
+      let body = { names: data.getNamesList() };
+
+      if (data.hasTensor()) {
+        data = data.getTensor();
+        body["tensor"] = {
+          shape: data.getShapeList(),
+          values: data.getValuesList()
+        };
+      } else {
+        body["ndarray"] = data.getNdarray();
+      }
+
+      let truth = call.request.getTruth();
+      let truth_data = truth.getData();
+      let truth_body = { names: truth_data.getNamesList() };
+
+      if (truth_data.hasTensor()) {
+        truth_data = truth_data.getTensor();
+        truth_body["tensor"] = {
+          shape: truth_data.getShapeList(),
+          values: truth_data.getValuesList()
+        };
+      } else {
+        truth_body["ndarray"] = truth_data.getNdarray();
+      }
+      result = send_feedback(
+        rest_data_to_array(body),
+        body.names,
+        rest_data_to_array(truth_body),
+        call.request.getReward()
+      );
+      callback(null, {});
+    }
+    var server = new grpc.Server();
+    server.addService(grpc_services.ModelService, { predict: predictEndpoint });
+    // server.addService(grpc_services.SeldonService, {
+    //   predict: predictEndpoint,
+    //   sendFeedback: feedbackEndpoint
+    // });
+    server.bind("0.0.0.0:" + port, grpc.ServerCredentials.createInsecure());
+    server.start();
+    console.log(`NodeJs GRPC Microservice listening on port ${port}!`);
+    function stopServer(code) {
+      server.forceShutdown();
+      console.log(`About to exit with code: ${code}`);
+    }
+    process.on("SIGINT", stopServer.bind(this));
+    process.on("SIGTERM", stopServer.bind(this));
+  }
+};
diff --git a/wrappers/s2i/nodejs/package.json b/wrappers/s2i/nodejs/package.json
index fd688cc3f6..8cd63e3093 100644
--- a/wrappers/s2i/nodejs/package.json
+++ b/wrappers/s2i/nodejs/package.json
@@ -15,6 +15,8 @@
     "argparse": "1.0.10",
     "body-parser": "1.18.3",
     "express": "4.16.3",
+    "google-protobuf": "3.6.1",
+    "grpc": "1.15.1",
     "numjs": "0.16.0"
   }
 }
diff --git a/wrappers/s2i/nodejs/prediction_grpc_pb.js b/wrappers/s2i/nodejs/prediction_grpc_pb.js
new file mode 100644
index 0000000000..afa057cf6f
--- /dev/null
+++ b/wrappers/s2i/nodejs/prediction_grpc_pb.js
@@ -0,0 +1,216 @@
+// GENERATED CODE -- DO NOT EDIT!
+
+'use strict';
+var grpc = require('grpc');
+var prediction_pb = require('./prediction_pb.js');
+var google_protobuf_struct_pb = require('google-protobuf/google/protobuf/struct_pb.js');
+
+function serialize_seldon_protos_Feedback(arg) {
+  if (!(arg instanceof prediction_pb.Feedback)) {
+    throw new Error('Expected argument of type seldon.protos.Feedback');
+  }
+  return new Buffer(arg.serializeBinary());
+}
+
+function deserialize_seldon_protos_Feedback(buffer_arg) {
+  return prediction_pb.Feedback.deserializeBinary(new Uint8Array(buffer_arg));
+}
+
+function serialize_seldon_protos_SeldonMessage(arg) {
+  if (!(arg instanceof prediction_pb.SeldonMessage)) {
+    throw new Error('Expected argument of type seldon.protos.SeldonMessage');
+  }
+  return new Buffer(arg.serializeBinary());
+}
+
+function deserialize_seldon_protos_SeldonMessage(buffer_arg) {
+  return prediction_pb.SeldonMessage.deserializeBinary(new Uint8Array(buffer_arg));
+}
+
+function serialize_seldon_protos_SeldonMessageList(arg) {
+  if (!(arg instanceof prediction_pb.SeldonMessageList)) {
+    throw new Error('Expected argument of type seldon.protos.SeldonMessageList');
+  }
+  return new Buffer(arg.serializeBinary());
+}
+
+function deserialize_seldon_protos_SeldonMessageList(buffer_arg) {
+  return prediction_pb.SeldonMessageList.deserializeBinary(new Uint8Array(buffer_arg));
+}
+
+
+// [END Messages]
+//
+// [START Services]
+//
+var GenericService = exports.GenericService = {
+  transformInput: {
+    path: '/seldon.protos.Generic/TransformInput',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.SeldonMessage,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_SeldonMessage,
+    requestDeserialize: deserialize_seldon_protos_SeldonMessage,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+  transformOutput: {
+    path: '/seldon.protos.Generic/TransformOutput',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.SeldonMessage,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_SeldonMessage,
+    requestDeserialize: deserialize_seldon_protos_SeldonMessage,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+  route: {
+    path: '/seldon.protos.Generic/Route',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.SeldonMessage,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_SeldonMessage,
+    requestDeserialize: deserialize_seldon_protos_SeldonMessage,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+  aggregate: {
+    path: '/seldon.protos.Generic/Aggregate',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.SeldonMessageList,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_SeldonMessageList,
+    requestDeserialize: deserialize_seldon_protos_SeldonMessageList,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+  sendFeedback: {
+    path: '/seldon.protos.Generic/SendFeedback',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.Feedback,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_Feedback,
+    requestDeserialize: deserialize_seldon_protos_Feedback,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+};
+
+exports.GenericClient = grpc.makeGenericClientConstructor(GenericService);
+var ModelService = exports.ModelService = {
+  predict: {
+    path: '/seldon.protos.Model/Predict',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.SeldonMessage,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_SeldonMessage,
+    requestDeserialize: deserialize_seldon_protos_SeldonMessage,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+};
+
+exports.ModelClient = grpc.makeGenericClientConstructor(ModelService);
+var RouterService = exports.RouterService = {
+  route: {
+    path: '/seldon.protos.Router/Route',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.SeldonMessage,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_SeldonMessage,
+    requestDeserialize: deserialize_seldon_protos_SeldonMessage,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+  sendFeedback: {
+    path: '/seldon.protos.Router/SendFeedback',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.Feedback,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_Feedback,
+    requestDeserialize: deserialize_seldon_protos_Feedback,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+};
+
+exports.RouterClient = grpc.makeGenericClientConstructor(RouterService);
+var TransformerService = exports.TransformerService = {
+  transformInput: {
+    path: '/seldon.protos.Transformer/TransformInput',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.SeldonMessage,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_SeldonMessage,
+    requestDeserialize: deserialize_seldon_protos_SeldonMessage,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+};
+
+exports.TransformerClient = grpc.makeGenericClientConstructor(TransformerService);
+var OutputTransformerService = exports.OutputTransformerService = {
+  transformOutput: {
+    path: '/seldon.protos.OutputTransformer/TransformOutput',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.SeldonMessage,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_SeldonMessage,
+    requestDeserialize: deserialize_seldon_protos_SeldonMessage,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+};
+
+exports.OutputTransformerClient = grpc.makeGenericClientConstructor(OutputTransformerService);
+var CombinerService = exports.CombinerService = {
+  aggregate: {
+    path: '/seldon.protos.Combiner/Aggregate',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.SeldonMessageList,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_SeldonMessageList,
+    requestDeserialize: deserialize_seldon_protos_SeldonMessageList,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+};
+
+exports.CombinerClient = grpc.makeGenericClientConstructor(CombinerService);
+var SeldonService = exports.SeldonService = {
+  predict: {
+    path: '/seldon.protos.Seldon/Predict',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.SeldonMessage,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_SeldonMessage,
+    requestDeserialize: deserialize_seldon_protos_SeldonMessage,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+  sendFeedback: {
+    path: '/seldon.protos.Seldon/SendFeedback',
+    requestStream: false,
+    responseStream: false,
+    requestType: prediction_pb.Feedback,
+    responseType: prediction_pb.SeldonMessage,
+    requestSerialize: serialize_seldon_protos_Feedback,
+    requestDeserialize: deserialize_seldon_protos_Feedback,
+    responseSerialize: serialize_seldon_protos_SeldonMessage,
+    responseDeserialize: deserialize_seldon_protos_SeldonMessage,
+  },
+};
+
+exports.SeldonClient = grpc.makeGenericClientConstructor(SeldonService);
diff --git a/wrappers/s2i/nodejs/prediction_pb.js b/wrappers/s2i/nodejs/prediction_pb.js
new file mode 100644
index 0000000000..40bfa71853
--- /dev/null
+++ b/wrappers/s2i/nodejs/prediction_pb.js
@@ -0,0 +1,1961 @@
+/**
+ * @fileoverview
+ * @enhanceable
+ * @suppress {messageConventions} JS Compiler reports an error if a variable or
+ *     field starts with 'MSG_' and isn't a translatable message.
+ * @public
+ */
+// GENERATED CODE -- DO NOT EDIT!
+
+var jspb = require('google-protobuf');
+var goog = jspb;
+var global = Function('return this')();
+
+var google_protobuf_struct_pb = require('google-protobuf/google/protobuf/struct_pb.js');
+goog.exportSymbol('proto.seldon.protos.DefaultData', null, global);
+goog.exportSymbol('proto.seldon.protos.Feedback', null, global);
+goog.exportSymbol('proto.seldon.protos.Meta', null, global);
+goog.exportSymbol('proto.seldon.protos.RequestResponse', null, global);
+goog.exportSymbol('proto.seldon.protos.SeldonMessage', null, global);
+goog.exportSymbol('proto.seldon.protos.SeldonMessageList', null, global);
+goog.exportSymbol('proto.seldon.protos.Status', null, global);
+goog.exportSymbol('proto.seldon.protos.Status.StatusFlag', null, global);
+goog.exportSymbol('proto.seldon.protos.Tensor', null, global);
+
+/**
+ * Generated by JsPbCodeGenerator.
+ * @param {Array=} opt_data Optional initial data array, typically from a
+ * server response, or constructed directly in Javascript. The array is used
+ * in place and becomes part of the constructed object. It is not cloned.
+ * If no data is provided, the constructed object will be empty, but still
+ * valid.
+ * @extends {jspb.Message}
+ * @constructor
+ */
+proto.seldon.protos.SeldonMessage = function(opt_data) {
+  jspb.Message.initialize(this, opt_data, 0, -1, null, proto.seldon.protos.SeldonMessage.oneofGroups_);
+};
+goog.inherits(proto.seldon.protos.SeldonMessage, jspb.Message);
+if (goog.DEBUG && !COMPILED) {
+  proto.seldon.protos.SeldonMessage.displayName = 'proto.seldon.protos.SeldonMessage';
+}
+/**
+ * Oneof group definitions for this message. Each group defines the field
+ * numbers belonging to that group. When of these fields' value is set, all
+ * other fields in the group are cleared. During deserialization, if multiple
+ * fields are encountered for a group, only the last value seen will be kept.
+ * @private {!Array<!Array<number>>}
+ * @const
+ */
+proto.seldon.protos.SeldonMessage.oneofGroups_ = [[3,4,5]];
+
+/**
+ * @enum {number}
+ */
+proto.seldon.protos.SeldonMessage.DataOneofCase = {
+  DATA_ONEOF_NOT_SET: 0,
+  DATA: 3,
+  BINDATA: 4,
+  STRDATA: 5
+};
+
+/**
+ * @return {proto.seldon.protos.SeldonMessage.DataOneofCase}
+ */
+proto.seldon.protos.SeldonMessage.prototype.getDataOneofCase = function() {
+  return /** @type {proto.seldon.protos.SeldonMessage.DataOneofCase} */(jspb.Message.computeOneofCase(this, proto.seldon.protos.SeldonMessage.oneofGroups_[0]));
+};
+
+
+
+if (jspb.Message.GENERATE_TO_OBJECT) {
+/**
+ * Creates an object representation of this proto suitable for use in Soy templates.
+ * Field names that are reserved in JavaScript and will be renamed to pb_name.
+ * To access a reserved field use, foo.pb_<name>, eg, foo.pb_default.
+ * For the list of reserved names please see:
+ *     com.google.apps.jspb.JsClassTemplate.JS_RESERVED_WORDS.
+ * @param {boolean=} opt_includeInstance Whether to include the JSPB instance
+ *     for transitional soy proto support: http://goto/soy-param-migration
+ * @return {!Object}
+ */
+proto.seldon.protos.SeldonMessage.prototype.toObject = function(opt_includeInstance) {
+  return proto.seldon.protos.SeldonMessage.toObject(opt_includeInstance, this);
+};
+
+
+/**
+ * Static version of the {@see toObject} method.
+ * @param {boolean|undefined} includeInstance Whether to include the JSPB
+ *     instance for transitional soy proto support:
+ *     http://goto/soy-param-migration
+ * @param {!proto.seldon.protos.SeldonMessage} msg The msg instance to transform.
+ * @return {!Object}
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.SeldonMessage.toObject = function(includeInstance, msg) {
+  var f, obj = {
+    status: (f = msg.getStatus()) && proto.seldon.protos.Status.toObject(includeInstance, f),
+    meta: (f = msg.getMeta()) && proto.seldon.protos.Meta.toObject(includeInstance, f),
+    data: (f = msg.getData()) && proto.seldon.protos.DefaultData.toObject(includeInstance, f),
+    bindata: msg.getBindata_asB64(),
+    strdata: jspb.Message.getFieldWithDefault(msg, 5, "")
+  };
+
+  if (includeInstance) {
+    obj.$jspbMessageInstance = msg;
+  }
+  return obj;
+};
+}
+
+
+/**
+ * Deserializes binary data (in protobuf wire format).
+ * @param {jspb.ByteSource} bytes The bytes to deserialize.
+ * @return {!proto.seldon.protos.SeldonMessage}
+ */
+proto.seldon.protos.SeldonMessage.deserializeBinary = function(bytes) {
+  var reader = new jspb.BinaryReader(bytes);
+  var msg = new proto.seldon.protos.SeldonMessage;
+  return proto.seldon.protos.SeldonMessage.deserializeBinaryFromReader(msg, reader);
+};
+
+
+/**
+ * Deserializes binary data (in protobuf wire format) from the
+ * given reader into the given message object.
+ * @param {!proto.seldon.protos.SeldonMessage} msg The message object to deserialize into.
+ * @param {!jspb.BinaryReader} reader The BinaryReader to use.
+ * @return {!proto.seldon.protos.SeldonMessage}
+ */
+proto.seldon.protos.SeldonMessage.deserializeBinaryFromReader = function(msg, reader) {
+  while (reader.nextField()) {
+    if (reader.isEndGroup()) {
+      break;
+    }
+    var field = reader.getFieldNumber();
+    switch (field) {
+    case 1:
+      var value = new proto.seldon.protos.Status;
+      reader.readMessage(value,proto.seldon.protos.Status.deserializeBinaryFromReader);
+      msg.setStatus(value);
+      break;
+    case 2:
+      var value = new proto.seldon.protos.Meta;
+      reader.readMessage(value,proto.seldon.protos.Meta.deserializeBinaryFromReader);
+      msg.setMeta(value);
+      break;
+    case 3:
+      var value = new proto.seldon.protos.DefaultData;
+      reader.readMessage(value,proto.seldon.protos.DefaultData.deserializeBinaryFromReader);
+      msg.setData(value);
+      break;
+    case 4:
+      var value = /** @type {!Uint8Array} */ (reader.readBytes());
+      msg.setBindata(value);
+      break;
+    case 5:
+      var value = /** @type {string} */ (reader.readString());
+      msg.setStrdata(value);
+      break;
+    default:
+      reader.skipField();
+      break;
+    }
+  }
+  return msg;
+};
+
+
+/**
+ * Serializes the message to binary data (in protobuf wire format).
+ * @return {!Uint8Array}
+ */
+proto.seldon.protos.SeldonMessage.prototype.serializeBinary = function() {
+  var writer = new jspb.BinaryWriter();
+  proto.seldon.protos.SeldonMessage.serializeBinaryToWriter(this, writer);
+  return writer.getResultBuffer();
+};
+
+
+/**
+ * Serializes the given message to binary data (in protobuf wire
+ * format), writing to the given BinaryWriter.
+ * @param {!proto.seldon.protos.SeldonMessage} message
+ * @param {!jspb.BinaryWriter} writer
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.SeldonMessage.serializeBinaryToWriter = function(message, writer) {
+  var f = undefined;
+  f = message.getStatus();
+  if (f != null) {
+    writer.writeMessage(
+      1,
+      f,
+      proto.seldon.protos.Status.serializeBinaryToWriter
+    );
+  }
+  f = message.getMeta();
+  if (f != null) {
+    writer.writeMessage(
+      2,
+      f,
+      proto.seldon.protos.Meta.serializeBinaryToWriter
+    );
+  }
+  f = message.getData();
+  if (f != null) {
+    writer.writeMessage(
+      3,
+      f,
+      proto.seldon.protos.DefaultData.serializeBinaryToWriter
+    );
+  }
+  f = /** @type {!(string|Uint8Array)} */ (jspb.Message.getField(message, 4));
+  if (f != null) {
+    writer.writeBytes(
+      4,
+      f
+    );
+  }
+  f = /** @type {string} */ (jspb.Message.getField(message, 5));
+  if (f != null) {
+    writer.writeString(
+      5,
+      f
+    );
+  }
+};
+
+
+/**
+ * optional Status status = 1;
+ * @return {?proto.seldon.protos.Status}
+ */
+proto.seldon.protos.SeldonMessage.prototype.getStatus = function() {
+  return /** @type{?proto.seldon.protos.Status} */ (
+    jspb.Message.getWrapperField(this, proto.seldon.protos.Status, 1));
+};
+
+
+/** @param {?proto.seldon.protos.Status|undefined} value */
+proto.seldon.protos.SeldonMessage.prototype.setStatus = function(value) {
+  jspb.Message.setWrapperField(this, 1, value);
+};
+
+
+proto.seldon.protos.SeldonMessage.prototype.clearStatus = function() {
+  this.setStatus(undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.SeldonMessage.prototype.hasStatus = function() {
+  return jspb.Message.getField(this, 1) != null;
+};
+
+
+/**
+ * optional Meta meta = 2;
+ * @return {?proto.seldon.protos.Meta}
+ */
+proto.seldon.protos.SeldonMessage.prototype.getMeta = function() {
+  return /** @type{?proto.seldon.protos.Meta} */ (
+    jspb.Message.getWrapperField(this, proto.seldon.protos.Meta, 2));
+};
+
+
+/** @param {?proto.seldon.protos.Meta|undefined} value */
+proto.seldon.protos.SeldonMessage.prototype.setMeta = function(value) {
+  jspb.Message.setWrapperField(this, 2, value);
+};
+
+
+proto.seldon.protos.SeldonMessage.prototype.clearMeta = function() {
+  this.setMeta(undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.SeldonMessage.prototype.hasMeta = function() {
+  return jspb.Message.getField(this, 2) != null;
+};
+
+
+/**
+ * optional DefaultData data = 3;
+ * @return {?proto.seldon.protos.DefaultData}
+ */
+proto.seldon.protos.SeldonMessage.prototype.getData = function() {
+  return /** @type{?proto.seldon.protos.DefaultData} */ (
+    jspb.Message.getWrapperField(this, proto.seldon.protos.DefaultData, 3));
+};
+
+
+/** @param {?proto.seldon.protos.DefaultData|undefined} value */
+proto.seldon.protos.SeldonMessage.prototype.setData = function(value) {
+  jspb.Message.setOneofWrapperField(this, 3, proto.seldon.protos.SeldonMessage.oneofGroups_[0], value);
+};
+
+
+proto.seldon.protos.SeldonMessage.prototype.clearData = function() {
+  this.setData(undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.SeldonMessage.prototype.hasData = function() {
+  return jspb.Message.getField(this, 3) != null;
+};
+
+
+/**
+ * optional bytes binData = 4;
+ * @return {!(string|Uint8Array)}
+ */
+proto.seldon.protos.SeldonMessage.prototype.getBindata = function() {
+  return /** @type {!(string|Uint8Array)} */ (jspb.Message.getFieldWithDefault(this, 4, ""));
+};
+
+
+/**
+ * optional bytes binData = 4;
+ * This is a type-conversion wrapper around `getBindata()`
+ * @return {string}
+ */
+proto.seldon.protos.SeldonMessage.prototype.getBindata_asB64 = function() {
+  return /** @type {string} */ (jspb.Message.bytesAsB64(
+      this.getBindata()));
+};
+
+
+/**
+ * optional bytes binData = 4;
+ * Note that Uint8Array is not supported on all browsers.
+ * @see http://caniuse.com/Uint8Array
+ * This is a type-conversion wrapper around `getBindata()`
+ * @return {!Uint8Array}
+ */
+proto.seldon.protos.SeldonMessage.prototype.getBindata_asU8 = function() {
+  return /** @type {!Uint8Array} */ (jspb.Message.bytesAsU8(
+      this.getBindata()));
+};
+
+
+/** @param {!(string|Uint8Array)} value */
+proto.seldon.protos.SeldonMessage.prototype.setBindata = function(value) {
+  jspb.Message.setOneofField(this, 4, proto.seldon.protos.SeldonMessage.oneofGroups_[0], value);
+};
+
+
+proto.seldon.protos.SeldonMessage.prototype.clearBindata = function() {
+  jspb.Message.setOneofField(this, 4, proto.seldon.protos.SeldonMessage.oneofGroups_[0], undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.SeldonMessage.prototype.hasBindata = function() {
+  return jspb.Message.getField(this, 4) != null;
+};
+
+
+/**
+ * optional string strData = 5;
+ * @return {string}
+ */
+proto.seldon.protos.SeldonMessage.prototype.getStrdata = function() {
+  return /** @type {string} */ (jspb.Message.getFieldWithDefault(this, 5, ""));
+};
+
+
+/** @param {string} value */
+proto.seldon.protos.SeldonMessage.prototype.setStrdata = function(value) {
+  jspb.Message.setOneofField(this, 5, proto.seldon.protos.SeldonMessage.oneofGroups_[0], value);
+};
+
+
+proto.seldon.protos.SeldonMessage.prototype.clearStrdata = function() {
+  jspb.Message.setOneofField(this, 5, proto.seldon.protos.SeldonMessage.oneofGroups_[0], undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.SeldonMessage.prototype.hasStrdata = function() {
+  return jspb.Message.getField(this, 5) != null;
+};
+
+
+
+/**
+ * Generated by JsPbCodeGenerator.
+ * @param {Array=} opt_data Optional initial data array, typically from a
+ * server response, or constructed directly in Javascript. The array is used
+ * in place and becomes part of the constructed object. It is not cloned.
+ * If no data is provided, the constructed object will be empty, but still
+ * valid.
+ * @extends {jspb.Message}
+ * @constructor
+ */
+proto.seldon.protos.DefaultData = function(opt_data) {
+  jspb.Message.initialize(this, opt_data, 0, -1, proto.seldon.protos.DefaultData.repeatedFields_, proto.seldon.protos.DefaultData.oneofGroups_);
+};
+goog.inherits(proto.seldon.protos.DefaultData, jspb.Message);
+if (goog.DEBUG && !COMPILED) {
+  proto.seldon.protos.DefaultData.displayName = 'proto.seldon.protos.DefaultData';
+}
+/**
+ * List of repeated fields within this message type.
+ * @private {!Array<number>}
+ * @const
+ */
+proto.seldon.protos.DefaultData.repeatedFields_ = [1];
+
+/**
+ * Oneof group definitions for this message. Each group defines the field
+ * numbers belonging to that group. When of these fields' value is set, all
+ * other fields in the group are cleared. During deserialization, if multiple
+ * fields are encountered for a group, only the last value seen will be kept.
+ * @private {!Array<!Array<number>>}
+ * @const
+ */
+proto.seldon.protos.DefaultData.oneofGroups_ = [[2,3]];
+
+/**
+ * @enum {number}
+ */
+proto.seldon.protos.DefaultData.DataOneofCase = {
+  DATA_ONEOF_NOT_SET: 0,
+  TENSOR: 2,
+  NDARRAY: 3
+};
+
+/**
+ * @return {proto.seldon.protos.DefaultData.DataOneofCase}
+ */
+proto.seldon.protos.DefaultData.prototype.getDataOneofCase = function() {
+  return /** @type {proto.seldon.protos.DefaultData.DataOneofCase} */(jspb.Message.computeOneofCase(this, proto.seldon.protos.DefaultData.oneofGroups_[0]));
+};
+
+
+
+if (jspb.Message.GENERATE_TO_OBJECT) {
+/**
+ * Creates an object representation of this proto suitable for use in Soy templates.
+ * Field names that are reserved in JavaScript and will be renamed to pb_name.
+ * To access a reserved field use, foo.pb_<name>, eg, foo.pb_default.
+ * For the list of reserved names please see:
+ *     com.google.apps.jspb.JsClassTemplate.JS_RESERVED_WORDS.
+ * @param {boolean=} opt_includeInstance Whether to include the JSPB instance
+ *     for transitional soy proto support: http://goto/soy-param-migration
+ * @return {!Object}
+ */
+proto.seldon.protos.DefaultData.prototype.toObject = function(opt_includeInstance) {
+  return proto.seldon.protos.DefaultData.toObject(opt_includeInstance, this);
+};
+
+
+/**
+ * Static version of the {@see toObject} method.
+ * @param {boolean|undefined} includeInstance Whether to include the JSPB
+ *     instance for transitional soy proto support:
+ *     http://goto/soy-param-migration
+ * @param {!proto.seldon.protos.DefaultData} msg The msg instance to transform.
+ * @return {!Object}
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.DefaultData.toObject = function(includeInstance, msg) {
+  var f, obj = {
+    namesList: jspb.Message.getRepeatedField(msg, 1),
+    tensor: (f = msg.getTensor()) && proto.seldon.protos.Tensor.toObject(includeInstance, f),
+    ndarray: (f = msg.getNdarray()) && google_protobuf_struct_pb.ListValue.toObject(includeInstance, f)
+  };
+
+  if (includeInstance) {
+    obj.$jspbMessageInstance = msg;
+  }
+  return obj;
+};
+}
+
+
+/**
+ * Deserializes binary data (in protobuf wire format).
+ * @param {jspb.ByteSource} bytes The bytes to deserialize.
+ * @return {!proto.seldon.protos.DefaultData}
+ */
+proto.seldon.protos.DefaultData.deserializeBinary = function(bytes) {
+  var reader = new jspb.BinaryReader(bytes);
+  var msg = new proto.seldon.protos.DefaultData;
+  return proto.seldon.protos.DefaultData.deserializeBinaryFromReader(msg, reader);
+};
+
+
+/**
+ * Deserializes binary data (in protobuf wire format) from the
+ * given reader into the given message object.
+ * @param {!proto.seldon.protos.DefaultData} msg The message object to deserialize into.
+ * @param {!jspb.BinaryReader} reader The BinaryReader to use.
+ * @return {!proto.seldon.protos.DefaultData}
+ */
+proto.seldon.protos.DefaultData.deserializeBinaryFromReader = function(msg, reader) {
+  while (reader.nextField()) {
+    if (reader.isEndGroup()) {
+      break;
+    }
+    var field = reader.getFieldNumber();
+    switch (field) {
+    case 1:
+      var value = /** @type {string} */ (reader.readString());
+      msg.addNames(value);
+      break;
+    case 2:
+      var value = new proto.seldon.protos.Tensor;
+      reader.readMessage(value,proto.seldon.protos.Tensor.deserializeBinaryFromReader);
+      msg.setTensor(value);
+      break;
+    case 3:
+      var value = new google_protobuf_struct_pb.ListValue;
+      reader.readMessage(value,google_protobuf_struct_pb.ListValue.deserializeBinaryFromReader);
+      msg.setNdarray(value);
+      break;
+    default:
+      reader.skipField();
+      break;
+    }
+  }
+  return msg;
+};
+
+
+/**
+ * Serializes the message to binary data (in protobuf wire format).
+ * @return {!Uint8Array}
+ */
+proto.seldon.protos.DefaultData.prototype.serializeBinary = function() {
+  var writer = new jspb.BinaryWriter();
+  proto.seldon.protos.DefaultData.serializeBinaryToWriter(this, writer);
+  return writer.getResultBuffer();
+};
+
+
+/**
+ * Serializes the given message to binary data (in protobuf wire
+ * format), writing to the given BinaryWriter.
+ * @param {!proto.seldon.protos.DefaultData} message
+ * @param {!jspb.BinaryWriter} writer
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.DefaultData.serializeBinaryToWriter = function(message, writer) {
+  var f = undefined;
+  f = message.getNamesList();
+  if (f.length > 0) {
+    writer.writeRepeatedString(
+      1,
+      f
+    );
+  }
+  f = message.getTensor();
+  if (f != null) {
+    writer.writeMessage(
+      2,
+      f,
+      proto.seldon.protos.Tensor.serializeBinaryToWriter
+    );
+  }
+  f = message.getNdarray();
+  if (f != null) {
+    writer.writeMessage(
+      3,
+      f,
+      google_protobuf_struct_pb.ListValue.serializeBinaryToWriter
+    );
+  }
+};
+
+
+/**
+ * repeated string names = 1;
+ * @return {!Array.<string>}
+ */
+proto.seldon.protos.DefaultData.prototype.getNamesList = function() {
+  return /** @type {!Array.<string>} */ (jspb.Message.getRepeatedField(this, 1));
+};
+
+
+/** @param {!Array.<string>} value */
+proto.seldon.protos.DefaultData.prototype.setNamesList = function(value) {
+  jspb.Message.setField(this, 1, value || []);
+};
+
+
+/**
+ * @param {!string} value
+ * @param {number=} opt_index
+ */
+proto.seldon.protos.DefaultData.prototype.addNames = function(value, opt_index) {
+  jspb.Message.addToRepeatedField(this, 1, value, opt_index);
+};
+
+
+proto.seldon.protos.DefaultData.prototype.clearNamesList = function() {
+  this.setNamesList([]);
+};
+
+
+/**
+ * optional Tensor tensor = 2;
+ * @return {?proto.seldon.protos.Tensor}
+ */
+proto.seldon.protos.DefaultData.prototype.getTensor = function() {
+  return /** @type{?proto.seldon.protos.Tensor} */ (
+    jspb.Message.getWrapperField(this, proto.seldon.protos.Tensor, 2));
+};
+
+
+/** @param {?proto.seldon.protos.Tensor|undefined} value */
+proto.seldon.protos.DefaultData.prototype.setTensor = function(value) {
+  jspb.Message.setOneofWrapperField(this, 2, proto.seldon.protos.DefaultData.oneofGroups_[0], value);
+};
+
+
+proto.seldon.protos.DefaultData.prototype.clearTensor = function() {
+  this.setTensor(undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.DefaultData.prototype.hasTensor = function() {
+  return jspb.Message.getField(this, 2) != null;
+};
+
+
+/**
+ * optional google.protobuf.ListValue ndarray = 3;
+ * @return {?proto.google.protobuf.ListValue}
+ */
+proto.seldon.protos.DefaultData.prototype.getNdarray = function() {
+  return /** @type{?proto.google.protobuf.ListValue} */ (
+    jspb.Message.getWrapperField(this, google_protobuf_struct_pb.ListValue, 3));
+};
+
+
+/** @param {?proto.google.protobuf.ListValue|undefined} value */
+proto.seldon.protos.DefaultData.prototype.setNdarray = function(value) {
+  jspb.Message.setOneofWrapperField(this, 3, proto.seldon.protos.DefaultData.oneofGroups_[0], value);
+};
+
+
+proto.seldon.protos.DefaultData.prototype.clearNdarray = function() {
+  this.setNdarray(undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.DefaultData.prototype.hasNdarray = function() {
+  return jspb.Message.getField(this, 3) != null;
+};
+
+
+
+/**
+ * Generated by JsPbCodeGenerator.
+ * @param {Array=} opt_data Optional initial data array, typically from a
+ * server response, or constructed directly in Javascript. The array is used
+ * in place and becomes part of the constructed object. It is not cloned.
+ * If no data is provided, the constructed object will be empty, but still
+ * valid.
+ * @extends {jspb.Message}
+ * @constructor
+ */
+proto.seldon.protos.Tensor = function(opt_data) {
+  jspb.Message.initialize(this, opt_data, 0, -1, proto.seldon.protos.Tensor.repeatedFields_, null);
+};
+goog.inherits(proto.seldon.protos.Tensor, jspb.Message);
+if (goog.DEBUG && !COMPILED) {
+  proto.seldon.protos.Tensor.displayName = 'proto.seldon.protos.Tensor';
+}
+/**
+ * List of repeated fields within this message type.
+ * @private {!Array<number>}
+ * @const
+ */
+proto.seldon.protos.Tensor.repeatedFields_ = [1,2];
+
+
+
+if (jspb.Message.GENERATE_TO_OBJECT) {
+/**
+ * Creates an object representation of this proto suitable for use in Soy templates.
+ * Field names that are reserved in JavaScript and will be renamed to pb_name.
+ * To access a reserved field use, foo.pb_<name>, eg, foo.pb_default.
+ * For the list of reserved names please see:
+ *     com.google.apps.jspb.JsClassTemplate.JS_RESERVED_WORDS.
+ * @param {boolean=} opt_includeInstance Whether to include the JSPB instance
+ *     for transitional soy proto support: http://goto/soy-param-migration
+ * @return {!Object}
+ */
+proto.seldon.protos.Tensor.prototype.toObject = function(opt_includeInstance) {
+  return proto.seldon.protos.Tensor.toObject(opt_includeInstance, this);
+};
+
+
+/**
+ * Static version of the {@see toObject} method.
+ * @param {boolean|undefined} includeInstance Whether to include the JSPB
+ *     instance for transitional soy proto support:
+ *     http://goto/soy-param-migration
+ * @param {!proto.seldon.protos.Tensor} msg The msg instance to transform.
+ * @return {!Object}
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.Tensor.toObject = function(includeInstance, msg) {
+  var f, obj = {
+    shapeList: jspb.Message.getRepeatedField(msg, 1),
+    valuesList: jspb.Message.getRepeatedFloatingPointField(msg, 2)
+  };
+
+  if (includeInstance) {
+    obj.$jspbMessageInstance = msg;
+  }
+  return obj;
+};
+}
+
+
+/**
+ * Deserializes binary data (in protobuf wire format).
+ * @param {jspb.ByteSource} bytes The bytes to deserialize.
+ * @return {!proto.seldon.protos.Tensor}
+ */
+proto.seldon.protos.Tensor.deserializeBinary = function(bytes) {
+  var reader = new jspb.BinaryReader(bytes);
+  var msg = new proto.seldon.protos.Tensor;
+  return proto.seldon.protos.Tensor.deserializeBinaryFromReader(msg, reader);
+};
+
+
+/**
+ * Deserializes binary data (in protobuf wire format) from the
+ * given reader into the given message object.
+ * @param {!proto.seldon.protos.Tensor} msg The message object to deserialize into.
+ * @param {!jspb.BinaryReader} reader The BinaryReader to use.
+ * @return {!proto.seldon.protos.Tensor}
+ */
+proto.seldon.protos.Tensor.deserializeBinaryFromReader = function(msg, reader) {
+  while (reader.nextField()) {
+    if (reader.isEndGroup()) {
+      break;
+    }
+    var field = reader.getFieldNumber();
+    switch (field) {
+    case 1:
+      var value = /** @type {!Array.<number>} */ (reader.readPackedInt32());
+      msg.setShapeList(value);
+      break;
+    case 2:
+      var value = /** @type {!Array.<number>} */ (reader.readPackedDouble());
+      msg.setValuesList(value);
+      break;
+    default:
+      reader.skipField();
+      break;
+    }
+  }
+  return msg;
+};
+
+
+/**
+ * Serializes the message to binary data (in protobuf wire format).
+ * @return {!Uint8Array}
+ */
+proto.seldon.protos.Tensor.prototype.serializeBinary = function() {
+  var writer = new jspb.BinaryWriter();
+  proto.seldon.protos.Tensor.serializeBinaryToWriter(this, writer);
+  return writer.getResultBuffer();
+};
+
+
+/**
+ * Serializes the given message to binary data (in protobuf wire
+ * format), writing to the given BinaryWriter.
+ * @param {!proto.seldon.protos.Tensor} message
+ * @param {!jspb.BinaryWriter} writer
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.Tensor.serializeBinaryToWriter = function(message, writer) {
+  var f = undefined;
+  f = message.getShapeList();
+  if (f.length > 0) {
+    writer.writePackedInt32(
+      1,
+      f
+    );
+  }
+  f = message.getValuesList();
+  if (f.length > 0) {
+    writer.writePackedDouble(
+      2,
+      f
+    );
+  }
+};
+
+
+/**
+ * repeated int32 shape = 1;
+ * @return {!Array.<number>}
+ */
+proto.seldon.protos.Tensor.prototype.getShapeList = function() {
+  return /** @type {!Array.<number>} */ (jspb.Message.getRepeatedField(this, 1));
+};
+
+
+/** @param {!Array.<number>} value */
+proto.seldon.protos.Tensor.prototype.setShapeList = function(value) {
+  jspb.Message.setField(this, 1, value || []);
+};
+
+
+/**
+ * @param {!number} value
+ * @param {number=} opt_index
+ */
+proto.seldon.protos.Tensor.prototype.addShape = function(value, opt_index) {
+  jspb.Message.addToRepeatedField(this, 1, value, opt_index);
+};
+
+
+proto.seldon.protos.Tensor.prototype.clearShapeList = function() {
+  this.setShapeList([]);
+};
+
+
+/**
+ * repeated double values = 2;
+ * @return {!Array.<number>}
+ */
+proto.seldon.protos.Tensor.prototype.getValuesList = function() {
+  return /** @type {!Array.<number>} */ (jspb.Message.getRepeatedFloatingPointField(this, 2));
+};
+
+
+/** @param {!Array.<number>} value */
+proto.seldon.protos.Tensor.prototype.setValuesList = function(value) {
+  jspb.Message.setField(this, 2, value || []);
+};
+
+
+/**
+ * @param {!number} value
+ * @param {number=} opt_index
+ */
+proto.seldon.protos.Tensor.prototype.addValues = function(value, opt_index) {
+  jspb.Message.addToRepeatedField(this, 2, value, opt_index);
+};
+
+
+proto.seldon.protos.Tensor.prototype.clearValuesList = function() {
+  this.setValuesList([]);
+};
+
+
+
+/**
+ * Generated by JsPbCodeGenerator.
+ * @param {Array=} opt_data Optional initial data array, typically from a
+ * server response, or constructed directly in Javascript. The array is used
+ * in place and becomes part of the constructed object. It is not cloned.
+ * If no data is provided, the constructed object will be empty, but still
+ * valid.
+ * @extends {jspb.Message}
+ * @constructor
+ */
+proto.seldon.protos.Meta = function(opt_data) {
+  jspb.Message.initialize(this, opt_data, 0, -1, null, null);
+};
+goog.inherits(proto.seldon.protos.Meta, jspb.Message);
+if (goog.DEBUG && !COMPILED) {
+  proto.seldon.protos.Meta.displayName = 'proto.seldon.protos.Meta';
+}
+
+
+if (jspb.Message.GENERATE_TO_OBJECT) {
+/**
+ * Creates an object representation of this proto suitable for use in Soy templates.
+ * Field names that are reserved in JavaScript and will be renamed to pb_name.
+ * To access a reserved field use, foo.pb_<name>, eg, foo.pb_default.
+ * For the list of reserved names please see:
+ *     com.google.apps.jspb.JsClassTemplate.JS_RESERVED_WORDS.
+ * @param {boolean=} opt_includeInstance Whether to include the JSPB instance
+ *     for transitional soy proto support: http://goto/soy-param-migration
+ * @return {!Object}
+ */
+proto.seldon.protos.Meta.prototype.toObject = function(opt_includeInstance) {
+  return proto.seldon.protos.Meta.toObject(opt_includeInstance, this);
+};
+
+
+/**
+ * Static version of the {@see toObject} method.
+ * @param {boolean|undefined} includeInstance Whether to include the JSPB
+ *     instance for transitional soy proto support:
+ *     http://goto/soy-param-migration
+ * @param {!proto.seldon.protos.Meta} msg The msg instance to transform.
+ * @return {!Object}
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.Meta.toObject = function(includeInstance, msg) {
+  var f, obj = {
+    puid: jspb.Message.getFieldWithDefault(msg, 1, ""),
+    tagsMap: (f = msg.getTagsMap()) ? f.toObject(includeInstance, proto.google.protobuf.Value.toObject) : [],
+    routingMap: (f = msg.getRoutingMap()) ? f.toObject(includeInstance, undefined) : []
+  };
+
+  if (includeInstance) {
+    obj.$jspbMessageInstance = msg;
+  }
+  return obj;
+};
+}
+
+
+/**
+ * Deserializes binary data (in protobuf wire format).
+ * @param {jspb.ByteSource} bytes The bytes to deserialize.
+ * @return {!proto.seldon.protos.Meta}
+ */
+proto.seldon.protos.Meta.deserializeBinary = function(bytes) {
+  var reader = new jspb.BinaryReader(bytes);
+  var msg = new proto.seldon.protos.Meta;
+  return proto.seldon.protos.Meta.deserializeBinaryFromReader(msg, reader);
+};
+
+
+/**
+ * Deserializes binary data (in protobuf wire format) from the
+ * given reader into the given message object.
+ * @param {!proto.seldon.protos.Meta} msg The message object to deserialize into.
+ * @param {!jspb.BinaryReader} reader The BinaryReader to use.
+ * @return {!proto.seldon.protos.Meta}
+ */
+proto.seldon.protos.Meta.deserializeBinaryFromReader = function(msg, reader) {
+  while (reader.nextField()) {
+    if (reader.isEndGroup()) {
+      break;
+    }
+    var field = reader.getFieldNumber();
+    switch (field) {
+    case 1:
+      var value = /** @type {string} */ (reader.readString());
+      msg.setPuid(value);
+      break;
+    case 2:
+      var value = msg.getTagsMap();
+      reader.readMessage(value, function(message, reader) {
+        jspb.Map.deserializeBinary(message, reader, jspb.BinaryReader.prototype.readString, jspb.BinaryReader.prototype.readMessage, proto.google.protobuf.Value.deserializeBinaryFromReader);
+         });
+      break;
+    case 3:
+      var value = msg.getRoutingMap();
+      reader.readMessage(value, function(message, reader) {
+        jspb.Map.deserializeBinary(message, reader, jspb.BinaryReader.prototype.readString, jspb.BinaryReader.prototype.readInt32);
+         });
+      break;
+    default:
+      reader.skipField();
+      break;
+    }
+  }
+  return msg;
+};
+
+
+/**
+ * Serializes the message to binary data (in protobuf wire format).
+ * @return {!Uint8Array}
+ */
+proto.seldon.protos.Meta.prototype.serializeBinary = function() {
+  var writer = new jspb.BinaryWriter();
+  proto.seldon.protos.Meta.serializeBinaryToWriter(this, writer);
+  return writer.getResultBuffer();
+};
+
+
+/**
+ * Serializes the given message to binary data (in protobuf wire
+ * format), writing to the given BinaryWriter.
+ * @param {!proto.seldon.protos.Meta} message
+ * @param {!jspb.BinaryWriter} writer
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.Meta.serializeBinaryToWriter = function(message, writer) {
+  var f = undefined;
+  f = message.getPuid();
+  if (f.length > 0) {
+    writer.writeString(
+      1,
+      f
+    );
+  }
+  f = message.getTagsMap(true);
+  if (f && f.getLength() > 0) {
+    f.serializeBinary(2, writer, jspb.BinaryWriter.prototype.writeString, jspb.BinaryWriter.prototype.writeMessage, proto.google.protobuf.Value.serializeBinaryToWriter);
+  }
+  f = message.getRoutingMap(true);
+  if (f && f.getLength() > 0) {
+    f.serializeBinary(3, writer, jspb.BinaryWriter.prototype.writeString, jspb.BinaryWriter.prototype.writeInt32);
+  }
+};
+
+
+/**
+ * optional string puid = 1;
+ * @return {string}
+ */
+proto.seldon.protos.Meta.prototype.getPuid = function() {
+  return /** @type {string} */ (jspb.Message.getFieldWithDefault(this, 1, ""));
+};
+
+
+/** @param {string} value */
+proto.seldon.protos.Meta.prototype.setPuid = function(value) {
+  jspb.Message.setField(this, 1, value);
+};
+
+
+/**
+ * map<string, google.protobuf.Value> tags = 2;
+ * @param {boolean=} opt_noLazyCreate Do not create the map if
+ * empty, instead returning `undefined`
+ * @return {!jspb.Map<string,!proto.google.protobuf.Value>}
+ */
+proto.seldon.protos.Meta.prototype.getTagsMap = function(opt_noLazyCreate) {
+  return /** @type {!jspb.Map<string,!proto.google.protobuf.Value>} */ (
+      jspb.Message.getMapField(this, 2, opt_noLazyCreate,
+      proto.google.protobuf.Value));
+};
+
+
+proto.seldon.protos.Meta.prototype.clearTagsMap = function() {
+  this.getTagsMap().clear();
+};
+
+
+/**
+ * map<string, int32> routing = 3;
+ * @param {boolean=} opt_noLazyCreate Do not create the map if
+ * empty, instead returning `undefined`
+ * @return {!jspb.Map<string,number>}
+ */
+proto.seldon.protos.Meta.prototype.getRoutingMap = function(opt_noLazyCreate) {
+  return /** @type {!jspb.Map<string,number>} */ (
+      jspb.Message.getMapField(this, 3, opt_noLazyCreate,
+      null));
+};
+
+
+proto.seldon.protos.Meta.prototype.clearRoutingMap = function() {
+  this.getRoutingMap().clear();
+};
+
+
+
+/**
+ * Generated by JsPbCodeGenerator.
+ * @param {Array=} opt_data Optional initial data array, typically from a
+ * server response, or constructed directly in Javascript. The array is used
+ * in place and becomes part of the constructed object. It is not cloned.
+ * If no data is provided, the constructed object will be empty, but still
+ * valid.
+ * @extends {jspb.Message}
+ * @constructor
+ */
+proto.seldon.protos.SeldonMessageList = function(opt_data) {
+  jspb.Message.initialize(this, opt_data, 0, -1, proto.seldon.protos.SeldonMessageList.repeatedFields_, null);
+};
+goog.inherits(proto.seldon.protos.SeldonMessageList, jspb.Message);
+if (goog.DEBUG && !COMPILED) {
+  proto.seldon.protos.SeldonMessageList.displayName = 'proto.seldon.protos.SeldonMessageList';
+}
+/**
+ * List of repeated fields within this message type.
+ * @private {!Array<number>}
+ * @const
+ */
+proto.seldon.protos.SeldonMessageList.repeatedFields_ = [1];
+
+
+
+if (jspb.Message.GENERATE_TO_OBJECT) {
+/**
+ * Creates an object representation of this proto suitable for use in Soy templates.
+ * Field names that are reserved in JavaScript and will be renamed to pb_name.
+ * To access a reserved field use, foo.pb_<name>, eg, foo.pb_default.
+ * For the list of reserved names please see:
+ *     com.google.apps.jspb.JsClassTemplate.JS_RESERVED_WORDS.
+ * @param {boolean=} opt_includeInstance Whether to include the JSPB instance
+ *     for transitional soy proto support: http://goto/soy-param-migration
+ * @return {!Object}
+ */
+proto.seldon.protos.SeldonMessageList.prototype.toObject = function(opt_includeInstance) {
+  return proto.seldon.protos.SeldonMessageList.toObject(opt_includeInstance, this);
+};
+
+
+/**
+ * Static version of the {@see toObject} method.
+ * @param {boolean|undefined} includeInstance Whether to include the JSPB
+ *     instance for transitional soy proto support:
+ *     http://goto/soy-param-migration
+ * @param {!proto.seldon.protos.SeldonMessageList} msg The msg instance to transform.
+ * @return {!Object}
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.SeldonMessageList.toObject = function(includeInstance, msg) {
+  var f, obj = {
+    seldonmessagesList: jspb.Message.toObjectList(msg.getSeldonmessagesList(),
+    proto.seldon.protos.SeldonMessage.toObject, includeInstance)
+  };
+
+  if (includeInstance) {
+    obj.$jspbMessageInstance = msg;
+  }
+  return obj;
+};
+}
+
+
+/**
+ * Deserializes binary data (in protobuf wire format).
+ * @param {jspb.ByteSource} bytes The bytes to deserialize.
+ * @return {!proto.seldon.protos.SeldonMessageList}
+ */
+proto.seldon.protos.SeldonMessageList.deserializeBinary = function(bytes) {
+  var reader = new jspb.BinaryReader(bytes);
+  var msg = new proto.seldon.protos.SeldonMessageList;
+  return proto.seldon.protos.SeldonMessageList.deserializeBinaryFromReader(msg, reader);
+};
+
+
+/**
+ * Deserializes binary data (in protobuf wire format) from the
+ * given reader into the given message object.
+ * @param {!proto.seldon.protos.SeldonMessageList} msg The message object to deserialize into.
+ * @param {!jspb.BinaryReader} reader The BinaryReader to use.
+ * @return {!proto.seldon.protos.SeldonMessageList}
+ */
+proto.seldon.protos.SeldonMessageList.deserializeBinaryFromReader = function(msg, reader) {
+  while (reader.nextField()) {
+    if (reader.isEndGroup()) {
+      break;
+    }
+    var field = reader.getFieldNumber();
+    switch (field) {
+    case 1:
+      var value = new proto.seldon.protos.SeldonMessage;
+      reader.readMessage(value,proto.seldon.protos.SeldonMessage.deserializeBinaryFromReader);
+      msg.addSeldonmessages(value);
+      break;
+    default:
+      reader.skipField();
+      break;
+    }
+  }
+  return msg;
+};
+
+
+/**
+ * Serializes the message to binary data (in protobuf wire format).
+ * @return {!Uint8Array}
+ */
+proto.seldon.protos.SeldonMessageList.prototype.serializeBinary = function() {
+  var writer = new jspb.BinaryWriter();
+  proto.seldon.protos.SeldonMessageList.serializeBinaryToWriter(this, writer);
+  return writer.getResultBuffer();
+};
+
+
+/**
+ * Serializes the given message to binary data (in protobuf wire
+ * format), writing to the given BinaryWriter.
+ * @param {!proto.seldon.protos.SeldonMessageList} message
+ * @param {!jspb.BinaryWriter} writer
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.SeldonMessageList.serializeBinaryToWriter = function(message, writer) {
+  var f = undefined;
+  f = message.getSeldonmessagesList();
+  if (f.length > 0) {
+    writer.writeRepeatedMessage(
+      1,
+      f,
+      proto.seldon.protos.SeldonMessage.serializeBinaryToWriter
+    );
+  }
+};
+
+
+/**
+ * repeated SeldonMessage seldonMessages = 1;
+ * @return {!Array.<!proto.seldon.protos.SeldonMessage>}
+ */
+proto.seldon.protos.SeldonMessageList.prototype.getSeldonmessagesList = function() {
+  return /** @type{!Array.<!proto.seldon.protos.SeldonMessage>} */ (
+    jspb.Message.getRepeatedWrapperField(this, proto.seldon.protos.SeldonMessage, 1));
+};
+
+
+/** @param {!Array.<!proto.seldon.protos.SeldonMessage>} value */
+proto.seldon.protos.SeldonMessageList.prototype.setSeldonmessagesList = function(value) {
+  jspb.Message.setRepeatedWrapperField(this, 1, value);
+};
+
+
+/**
+ * @param {!proto.seldon.protos.SeldonMessage=} opt_value
+ * @param {number=} opt_index
+ * @return {!proto.seldon.protos.SeldonMessage}
+ */
+proto.seldon.protos.SeldonMessageList.prototype.addSeldonmessages = function(opt_value, opt_index) {
+  return jspb.Message.addToRepeatedWrapperField(this, 1, opt_value, proto.seldon.protos.SeldonMessage, opt_index);
+};
+
+
+proto.seldon.protos.SeldonMessageList.prototype.clearSeldonmessagesList = function() {
+  this.setSeldonmessagesList([]);
+};
+
+
+
+/**
+ * Generated by JsPbCodeGenerator.
+ * @param {Array=} opt_data Optional initial data array, typically from a
+ * server response, or constructed directly in Javascript. The array is used
+ * in place and becomes part of the constructed object. It is not cloned.
+ * If no data is provided, the constructed object will be empty, but still
+ * valid.
+ * @extends {jspb.Message}
+ * @constructor
+ */
+proto.seldon.protos.Status = function(opt_data) {
+  jspb.Message.initialize(this, opt_data, 0, -1, null, null);
+};
+goog.inherits(proto.seldon.protos.Status, jspb.Message);
+if (goog.DEBUG && !COMPILED) {
+  proto.seldon.protos.Status.displayName = 'proto.seldon.protos.Status';
+}
+
+
+if (jspb.Message.GENERATE_TO_OBJECT) {
+/**
+ * Creates an object representation of this proto suitable for use in Soy templates.
+ * Field names that are reserved in JavaScript and will be renamed to pb_name.
+ * To access a reserved field use, foo.pb_<name>, eg, foo.pb_default.
+ * For the list of reserved names please see:
+ *     com.google.apps.jspb.JsClassTemplate.JS_RESERVED_WORDS.
+ * @param {boolean=} opt_includeInstance Whether to include the JSPB instance
+ *     for transitional soy proto support: http://goto/soy-param-migration
+ * @return {!Object}
+ */
+proto.seldon.protos.Status.prototype.toObject = function(opt_includeInstance) {
+  return proto.seldon.protos.Status.toObject(opt_includeInstance, this);
+};
+
+
+/**
+ * Static version of the {@see toObject} method.
+ * @param {boolean|undefined} includeInstance Whether to include the JSPB
+ *     instance for transitional soy proto support:
+ *     http://goto/soy-param-migration
+ * @param {!proto.seldon.protos.Status} msg The msg instance to transform.
+ * @return {!Object}
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.Status.toObject = function(includeInstance, msg) {
+  var f, obj = {
+    code: jspb.Message.getFieldWithDefault(msg, 1, 0),
+    info: jspb.Message.getFieldWithDefault(msg, 2, ""),
+    reason: jspb.Message.getFieldWithDefault(msg, 3, ""),
+    status: jspb.Message.getFieldWithDefault(msg, 4, 0)
+  };
+
+  if (includeInstance) {
+    obj.$jspbMessageInstance = msg;
+  }
+  return obj;
+};
+}
+
+
+/**
+ * Deserializes binary data (in protobuf wire format).
+ * @param {jspb.ByteSource} bytes The bytes to deserialize.
+ * @return {!proto.seldon.protos.Status}
+ */
+proto.seldon.protos.Status.deserializeBinary = function(bytes) {
+  var reader = new jspb.BinaryReader(bytes);
+  var msg = new proto.seldon.protos.Status;
+  return proto.seldon.protos.Status.deserializeBinaryFromReader(msg, reader);
+};
+
+
+/**
+ * Deserializes binary data (in protobuf wire format) from the
+ * given reader into the given message object.
+ * @param {!proto.seldon.protos.Status} msg The message object to deserialize into.
+ * @param {!jspb.BinaryReader} reader The BinaryReader to use.
+ * @return {!proto.seldon.protos.Status}
+ */
+proto.seldon.protos.Status.deserializeBinaryFromReader = function(msg, reader) {
+  while (reader.nextField()) {
+    if (reader.isEndGroup()) {
+      break;
+    }
+    var field = reader.getFieldNumber();
+    switch (field) {
+    case 1:
+      var value = /** @type {number} */ (reader.readInt32());
+      msg.setCode(value);
+      break;
+    case 2:
+      var value = /** @type {string} */ (reader.readString());
+      msg.setInfo(value);
+      break;
+    case 3:
+      var value = /** @type {string} */ (reader.readString());
+      msg.setReason(value);
+      break;
+    case 4:
+      var value = /** @type {!proto.seldon.protos.Status.StatusFlag} */ (reader.readEnum());
+      msg.setStatus(value);
+      break;
+    default:
+      reader.skipField();
+      break;
+    }
+  }
+  return msg;
+};
+
+
+/**
+ * Serializes the message to binary data (in protobuf wire format).
+ * @return {!Uint8Array}
+ */
+proto.seldon.protos.Status.prototype.serializeBinary = function() {
+  var writer = new jspb.BinaryWriter();
+  proto.seldon.protos.Status.serializeBinaryToWriter(this, writer);
+  return writer.getResultBuffer();
+};
+
+
+/**
+ * Serializes the given message to binary data (in protobuf wire
+ * format), writing to the given BinaryWriter.
+ * @param {!proto.seldon.protos.Status} message
+ * @param {!jspb.BinaryWriter} writer
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.Status.serializeBinaryToWriter = function(message, writer) {
+  var f = undefined;
+  f = message.getCode();
+  if (f !== 0) {
+    writer.writeInt32(
+      1,
+      f
+    );
+  }
+  f = message.getInfo();
+  if (f.length > 0) {
+    writer.writeString(
+      2,
+      f
+    );
+  }
+  f = message.getReason();
+  if (f.length > 0) {
+    writer.writeString(
+      3,
+      f
+    );
+  }
+  f = message.getStatus();
+  if (f !== 0.0) {
+    writer.writeEnum(
+      4,
+      f
+    );
+  }
+};
+
+
+/**
+ * @enum {number}
+ */
+proto.seldon.protos.Status.StatusFlag = {
+  SUCCESS: 0,
+  FAILURE: 1
+};
+
+/**
+ * optional int32 code = 1;
+ * @return {number}
+ */
+proto.seldon.protos.Status.prototype.getCode = function() {
+  return /** @type {number} */ (jspb.Message.getFieldWithDefault(this, 1, 0));
+};
+
+
+/** @param {number} value */
+proto.seldon.protos.Status.prototype.setCode = function(value) {
+  jspb.Message.setField(this, 1, value);
+};
+
+
+/**
+ * optional string info = 2;
+ * @return {string}
+ */
+proto.seldon.protos.Status.prototype.getInfo = function() {
+  return /** @type {string} */ (jspb.Message.getFieldWithDefault(this, 2, ""));
+};
+
+
+/** @param {string} value */
+proto.seldon.protos.Status.prototype.setInfo = function(value) {
+  jspb.Message.setField(this, 2, value);
+};
+
+
+/**
+ * optional string reason = 3;
+ * @return {string}
+ */
+proto.seldon.protos.Status.prototype.getReason = function() {
+  return /** @type {string} */ (jspb.Message.getFieldWithDefault(this, 3, ""));
+};
+
+
+/** @param {string} value */
+proto.seldon.protos.Status.prototype.setReason = function(value) {
+  jspb.Message.setField(this, 3, value);
+};
+
+
+/**
+ * optional StatusFlag status = 4;
+ * @return {!proto.seldon.protos.Status.StatusFlag}
+ */
+proto.seldon.protos.Status.prototype.getStatus = function() {
+  return /** @type {!proto.seldon.protos.Status.StatusFlag} */ (jspb.Message.getFieldWithDefault(this, 4, 0));
+};
+
+
+/** @param {!proto.seldon.protos.Status.StatusFlag} value */
+proto.seldon.protos.Status.prototype.setStatus = function(value) {
+  jspb.Message.setField(this, 4, value);
+};
+
+
+
+/**
+ * Generated by JsPbCodeGenerator.
+ * @param {Array=} opt_data Optional initial data array, typically from a
+ * server response, or constructed directly in Javascript. The array is used
+ * in place and becomes part of the constructed object. It is not cloned.
+ * If no data is provided, the constructed object will be empty, but still
+ * valid.
+ * @extends {jspb.Message}
+ * @constructor
+ */
+proto.seldon.protos.Feedback = function(opt_data) {
+  jspb.Message.initialize(this, opt_data, 0, -1, null, null);
+};
+goog.inherits(proto.seldon.protos.Feedback, jspb.Message);
+if (goog.DEBUG && !COMPILED) {
+  proto.seldon.protos.Feedback.displayName = 'proto.seldon.protos.Feedback';
+}
+
+
+if (jspb.Message.GENERATE_TO_OBJECT) {
+/**
+ * Creates an object representation of this proto suitable for use in Soy templates.
+ * Field names that are reserved in JavaScript and will be renamed to pb_name.
+ * To access a reserved field use, foo.pb_<name>, eg, foo.pb_default.
+ * For the list of reserved names please see:
+ *     com.google.apps.jspb.JsClassTemplate.JS_RESERVED_WORDS.
+ * @param {boolean=} opt_includeInstance Whether to include the JSPB instance
+ *     for transitional soy proto support: http://goto/soy-param-migration
+ * @return {!Object}
+ */
+proto.seldon.protos.Feedback.prototype.toObject = function(opt_includeInstance) {
+  return proto.seldon.protos.Feedback.toObject(opt_includeInstance, this);
+};
+
+
+/**
+ * Static version of the {@see toObject} method.
+ * @param {boolean|undefined} includeInstance Whether to include the JSPB
+ *     instance for transitional soy proto support:
+ *     http://goto/soy-param-migration
+ * @param {!proto.seldon.protos.Feedback} msg The msg instance to transform.
+ * @return {!Object}
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.Feedback.toObject = function(includeInstance, msg) {
+  var f, obj = {
+    request: (f = msg.getRequest()) && proto.seldon.protos.SeldonMessage.toObject(includeInstance, f),
+    response: (f = msg.getResponse()) && proto.seldon.protos.SeldonMessage.toObject(includeInstance, f),
+    reward: +jspb.Message.getFieldWithDefault(msg, 3, 0.0),
+    truth: (f = msg.getTruth()) && proto.seldon.protos.SeldonMessage.toObject(includeInstance, f)
+  };
+
+  if (includeInstance) {
+    obj.$jspbMessageInstance = msg;
+  }
+  return obj;
+};
+}
+
+
+/**
+ * Deserializes binary data (in protobuf wire format).
+ * @param {jspb.ByteSource} bytes The bytes to deserialize.
+ * @return {!proto.seldon.protos.Feedback}
+ */
+proto.seldon.protos.Feedback.deserializeBinary = function(bytes) {
+  var reader = new jspb.BinaryReader(bytes);
+  var msg = new proto.seldon.protos.Feedback;
+  return proto.seldon.protos.Feedback.deserializeBinaryFromReader(msg, reader);
+};
+
+
+/**
+ * Deserializes binary data (in protobuf wire format) from the
+ * given reader into the given message object.
+ * @param {!proto.seldon.protos.Feedback} msg The message object to deserialize into.
+ * @param {!jspb.BinaryReader} reader The BinaryReader to use.
+ * @return {!proto.seldon.protos.Feedback}
+ */
+proto.seldon.protos.Feedback.deserializeBinaryFromReader = function(msg, reader) {
+  while (reader.nextField()) {
+    if (reader.isEndGroup()) {
+      break;
+    }
+    var field = reader.getFieldNumber();
+    switch (field) {
+    case 1:
+      var value = new proto.seldon.protos.SeldonMessage;
+      reader.readMessage(value,proto.seldon.protos.SeldonMessage.deserializeBinaryFromReader);
+      msg.setRequest(value);
+      break;
+    case 2:
+      var value = new proto.seldon.protos.SeldonMessage;
+      reader.readMessage(value,proto.seldon.protos.SeldonMessage.deserializeBinaryFromReader);
+      msg.setResponse(value);
+      break;
+    case 3:
+      var value = /** @type {number} */ (reader.readFloat());
+      msg.setReward(value);
+      break;
+    case 4:
+      var value = new proto.seldon.protos.SeldonMessage;
+      reader.readMessage(value,proto.seldon.protos.SeldonMessage.deserializeBinaryFromReader);
+      msg.setTruth(value);
+      break;
+    default:
+      reader.skipField();
+      break;
+    }
+  }
+  return msg;
+};
+
+
+/**
+ * Serializes the message to binary data (in protobuf wire format).
+ * @return {!Uint8Array}
+ */
+proto.seldon.protos.Feedback.prototype.serializeBinary = function() {
+  var writer = new jspb.BinaryWriter();
+  proto.seldon.protos.Feedback.serializeBinaryToWriter(this, writer);
+  return writer.getResultBuffer();
+};
+
+
+/**
+ * Serializes the given message to binary data (in protobuf wire
+ * format), writing to the given BinaryWriter.
+ * @param {!proto.seldon.protos.Feedback} message
+ * @param {!jspb.BinaryWriter} writer
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.Feedback.serializeBinaryToWriter = function(message, writer) {
+  var f = undefined;
+  f = message.getRequest();
+  if (f != null) {
+    writer.writeMessage(
+      1,
+      f,
+      proto.seldon.protos.SeldonMessage.serializeBinaryToWriter
+    );
+  }
+  f = message.getResponse();
+  if (f != null) {
+    writer.writeMessage(
+      2,
+      f,
+      proto.seldon.protos.SeldonMessage.serializeBinaryToWriter
+    );
+  }
+  f = message.getReward();
+  if (f !== 0.0) {
+    writer.writeFloat(
+      3,
+      f
+    );
+  }
+  f = message.getTruth();
+  if (f != null) {
+    writer.writeMessage(
+      4,
+      f,
+      proto.seldon.protos.SeldonMessage.serializeBinaryToWriter
+    );
+  }
+};
+
+
+/**
+ * optional SeldonMessage request = 1;
+ * @return {?proto.seldon.protos.SeldonMessage}
+ */
+proto.seldon.protos.Feedback.prototype.getRequest = function() {
+  return /** @type{?proto.seldon.protos.SeldonMessage} */ (
+    jspb.Message.getWrapperField(this, proto.seldon.protos.SeldonMessage, 1));
+};
+
+
+/** @param {?proto.seldon.protos.SeldonMessage|undefined} value */
+proto.seldon.protos.Feedback.prototype.setRequest = function(value) {
+  jspb.Message.setWrapperField(this, 1, value);
+};
+
+
+proto.seldon.protos.Feedback.prototype.clearRequest = function() {
+  this.setRequest(undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.Feedback.prototype.hasRequest = function() {
+  return jspb.Message.getField(this, 1) != null;
+};
+
+
+/**
+ * optional SeldonMessage response = 2;
+ * @return {?proto.seldon.protos.SeldonMessage}
+ */
+proto.seldon.protos.Feedback.prototype.getResponse = function() {
+  return /** @type{?proto.seldon.protos.SeldonMessage} */ (
+    jspb.Message.getWrapperField(this, proto.seldon.protos.SeldonMessage, 2));
+};
+
+
+/** @param {?proto.seldon.protos.SeldonMessage|undefined} value */
+proto.seldon.protos.Feedback.prototype.setResponse = function(value) {
+  jspb.Message.setWrapperField(this, 2, value);
+};
+
+
+proto.seldon.protos.Feedback.prototype.clearResponse = function() {
+  this.setResponse(undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.Feedback.prototype.hasResponse = function() {
+  return jspb.Message.getField(this, 2) != null;
+};
+
+
+/**
+ * optional float reward = 3;
+ * @return {number}
+ */
+proto.seldon.protos.Feedback.prototype.getReward = function() {
+  return /** @type {number} */ (+jspb.Message.getFieldWithDefault(this, 3, 0.0));
+};
+
+
+/** @param {number} value */
+proto.seldon.protos.Feedback.prototype.setReward = function(value) {
+  jspb.Message.setField(this, 3, value);
+};
+
+
+/**
+ * optional SeldonMessage truth = 4;
+ * @return {?proto.seldon.protos.SeldonMessage}
+ */
+proto.seldon.protos.Feedback.prototype.getTruth = function() {
+  return /** @type{?proto.seldon.protos.SeldonMessage} */ (
+    jspb.Message.getWrapperField(this, proto.seldon.protos.SeldonMessage, 4));
+};
+
+
+/** @param {?proto.seldon.protos.SeldonMessage|undefined} value */
+proto.seldon.protos.Feedback.prototype.setTruth = function(value) {
+  jspb.Message.setWrapperField(this, 4, value);
+};
+
+
+proto.seldon.protos.Feedback.prototype.clearTruth = function() {
+  this.setTruth(undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.Feedback.prototype.hasTruth = function() {
+  return jspb.Message.getField(this, 4) != null;
+};
+
+
+
+/**
+ * Generated by JsPbCodeGenerator.
+ * @param {Array=} opt_data Optional initial data array, typically from a
+ * server response, or constructed directly in Javascript. The array is used
+ * in place and becomes part of the constructed object. It is not cloned.
+ * If no data is provided, the constructed object will be empty, but still
+ * valid.
+ * @extends {jspb.Message}
+ * @constructor
+ */
+proto.seldon.protos.RequestResponse = function(opt_data) {
+  jspb.Message.initialize(this, opt_data, 0, -1, null, null);
+};
+goog.inherits(proto.seldon.protos.RequestResponse, jspb.Message);
+if (goog.DEBUG && !COMPILED) {
+  proto.seldon.protos.RequestResponse.displayName = 'proto.seldon.protos.RequestResponse';
+}
+
+
+if (jspb.Message.GENERATE_TO_OBJECT) {
+/**
+ * Creates an object representation of this proto suitable for use in Soy templates.
+ * Field names that are reserved in JavaScript and will be renamed to pb_name.
+ * To access a reserved field use, foo.pb_<name>, eg, foo.pb_default.
+ * For the list of reserved names please see:
+ *     com.google.apps.jspb.JsClassTemplate.JS_RESERVED_WORDS.
+ * @param {boolean=} opt_includeInstance Whether to include the JSPB instance
+ *     for transitional soy proto support: http://goto/soy-param-migration
+ * @return {!Object}
+ */
+proto.seldon.protos.RequestResponse.prototype.toObject = function(opt_includeInstance) {
+  return proto.seldon.protos.RequestResponse.toObject(opt_includeInstance, this);
+};
+
+
+/**
+ * Static version of the {@see toObject} method.
+ * @param {boolean|undefined} includeInstance Whether to include the JSPB
+ *     instance for transitional soy proto support:
+ *     http://goto/soy-param-migration
+ * @param {!proto.seldon.protos.RequestResponse} msg The msg instance to transform.
+ * @return {!Object}
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.RequestResponse.toObject = function(includeInstance, msg) {
+  var f, obj = {
+    request: (f = msg.getRequest()) && proto.seldon.protos.SeldonMessage.toObject(includeInstance, f),
+    response: (f = msg.getResponse()) && proto.seldon.protos.SeldonMessage.toObject(includeInstance, f)
+  };
+
+  if (includeInstance) {
+    obj.$jspbMessageInstance = msg;
+  }
+  return obj;
+};
+}
+
+
+/**
+ * Deserializes binary data (in protobuf wire format).
+ * @param {jspb.ByteSource} bytes The bytes to deserialize.
+ * @return {!proto.seldon.protos.RequestResponse}
+ */
+proto.seldon.protos.RequestResponse.deserializeBinary = function(bytes) {
+  var reader = new jspb.BinaryReader(bytes);
+  var msg = new proto.seldon.protos.RequestResponse;
+  return proto.seldon.protos.RequestResponse.deserializeBinaryFromReader(msg, reader);
+};
+
+
+/**
+ * Deserializes binary data (in protobuf wire format) from the
+ * given reader into the given message object.
+ * @param {!proto.seldon.protos.RequestResponse} msg The message object to deserialize into.
+ * @param {!jspb.BinaryReader} reader The BinaryReader to use.
+ * @return {!proto.seldon.protos.RequestResponse}
+ */
+proto.seldon.protos.RequestResponse.deserializeBinaryFromReader = function(msg, reader) {
+  while (reader.nextField()) {
+    if (reader.isEndGroup()) {
+      break;
+    }
+    var field = reader.getFieldNumber();
+    switch (field) {
+    case 1:
+      var value = new proto.seldon.protos.SeldonMessage;
+      reader.readMessage(value,proto.seldon.protos.SeldonMessage.deserializeBinaryFromReader);
+      msg.setRequest(value);
+      break;
+    case 2:
+      var value = new proto.seldon.protos.SeldonMessage;
+      reader.readMessage(value,proto.seldon.protos.SeldonMessage.deserializeBinaryFromReader);
+      msg.setResponse(value);
+      break;
+    default:
+      reader.skipField();
+      break;
+    }
+  }
+  return msg;
+};
+
+
+/**
+ * Serializes the message to binary data (in protobuf wire format).
+ * @return {!Uint8Array}
+ */
+proto.seldon.protos.RequestResponse.prototype.serializeBinary = function() {
+  var writer = new jspb.BinaryWriter();
+  proto.seldon.protos.RequestResponse.serializeBinaryToWriter(this, writer);
+  return writer.getResultBuffer();
+};
+
+
+/**
+ * Serializes the given message to binary data (in protobuf wire
+ * format), writing to the given BinaryWriter.
+ * @param {!proto.seldon.protos.RequestResponse} message
+ * @param {!jspb.BinaryWriter} writer
+ * @suppress {unusedLocalVariables} f is only used for nested messages
+ */
+proto.seldon.protos.RequestResponse.serializeBinaryToWriter = function(message, writer) {
+  var f = undefined;
+  f = message.getRequest();
+  if (f != null) {
+    writer.writeMessage(
+      1,
+      f,
+      proto.seldon.protos.SeldonMessage.serializeBinaryToWriter
+    );
+  }
+  f = message.getResponse();
+  if (f != null) {
+    writer.writeMessage(
+      2,
+      f,
+      proto.seldon.protos.SeldonMessage.serializeBinaryToWriter
+    );
+  }
+};
+
+
+/**
+ * optional SeldonMessage request = 1;
+ * @return {?proto.seldon.protos.SeldonMessage}
+ */
+proto.seldon.protos.RequestResponse.prototype.getRequest = function() {
+  return /** @type{?proto.seldon.protos.SeldonMessage} */ (
+    jspb.Message.getWrapperField(this, proto.seldon.protos.SeldonMessage, 1));
+};
+
+
+/** @param {?proto.seldon.protos.SeldonMessage|undefined} value */
+proto.seldon.protos.RequestResponse.prototype.setRequest = function(value) {
+  jspb.Message.setWrapperField(this, 1, value);
+};
+
+
+proto.seldon.protos.RequestResponse.prototype.clearRequest = function() {
+  this.setRequest(undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.RequestResponse.prototype.hasRequest = function() {
+  return jspb.Message.getField(this, 1) != null;
+};
+
+
+/**
+ * optional SeldonMessage response = 2;
+ * @return {?proto.seldon.protos.SeldonMessage}
+ */
+proto.seldon.protos.RequestResponse.prototype.getResponse = function() {
+  return /** @type{?proto.seldon.protos.SeldonMessage} */ (
+    jspb.Message.getWrapperField(this, proto.seldon.protos.SeldonMessage, 2));
+};
+
+
+/** @param {?proto.seldon.protos.SeldonMessage|undefined} value */
+proto.seldon.protos.RequestResponse.prototype.setResponse = function(value) {
+  jspb.Message.setWrapperField(this, 2, value);
+};
+
+
+proto.seldon.protos.RequestResponse.prototype.clearResponse = function() {
+  this.setResponse(undefined);
+};
+
+
+/**
+ * Returns whether this field is set.
+ * @return {!boolean}
+ */
+proto.seldon.protos.RequestResponse.prototype.hasResponse = function() {
+  return jspb.Message.getField(this, 2) != null;
+};
+
+
+goog.object.extend(exports, proto.seldon.protos);
diff --git a/wrappers/s2i/nodejs/test/model-template-app/.s2i/environment b/wrappers/s2i/nodejs/test/model-template-app/.s2i/environment
index afcae69f8c..aab3ea3f7f 100644
--- a/wrappers/s2i/nodejs/test/model-template-app/.s2i/environment
+++ b/wrappers/s2i/nodejs/test/model-template-app/.s2i/environment
@@ -1,4 +1,4 @@
 MODEL_NAME=MyModel.js
 API_TYPE=REST
 SERVICE_TYPE=MODEL
-PERSISTENCE=0
+PERSISTENCE=0
\ No newline at end of file
diff --git a/wrappers/s2i/nodejs/test/model-template-app/README.md b/wrappers/s2i/nodejs/test/model-template-app/README.md
index d4033b8e73..9758a68243 100644
--- a/wrappers/s2i/nodejs/test/model-template-app/README.md
+++ b/wrappers/s2i/nodejs/test/model-template-app/README.md
@@ -1,4 +1,4 @@
-## NodeJs tensorflow example
+## NodeJs model example
 
 This model example takes an input of 10 different features and predicts a out for the same.
 For the training part it uses a random normally distributed input set of 100 rows i.e a data set of [100,10] and trains it for another random normally distributed data set of size [100,1].
diff --git a/wrappers/s2i/nodejs/test/model-template-app/package.json b/wrappers/s2i/nodejs/test/model-template-app/package.json
index eb9fe2383e..ae1e3493cf 100644
--- a/wrappers/s2i/nodejs/test/model-template-app/package.json
+++ b/wrappers/s2i/nodejs/test/model-template-app/package.json
@@ -1,5 +1,5 @@
 {
-  "name": "node_tensorflow",
+  "name": "model-template-app",
   "version": "1.0.0",
   "description": "",
   "main": "predict.js",
diff --git a/wrappers/s2i/nodejs/test/run b/wrappers/s2i/nodejs/test/run
index c69bcd4f9c..ac95bc98c4 100755
--- a/wrappers/s2i/nodejs/test/run
+++ b/wrappers/s2i/nodejs/test/run
@@ -171,7 +171,7 @@ test_feedback() {
 
 # Build the application image twice to ensure the 'save-artifacts' and
 # 'restore-artifacts' scripts are working properly
-array=( 'model' )
+array=( 'model' 'transformer' )
 for i in "${array[@]}"
 do
     cid_file=$($MKTEMP_EXEC -u --suffix=.cid)
diff --git a/wrappers/s2i/nodejs/test/transformer-template-app/.s2i/environment b/wrappers/s2i/nodejs/test/transformer-template-app/.s2i/environment
new file mode 100644
index 0000000000..6bb7e6ac4b
--- /dev/null
+++ b/wrappers/s2i/nodejs/test/transformer-template-app/.s2i/environment
@@ -0,0 +1,4 @@
+MODEL_NAME=MyTransformer.js
+API_TYPE=REST
+SERVICE_TYPE=TRANSFORMER
+PERSISTENCE=0
\ No newline at end of file
diff --git a/wrappers/s2i/nodejs/test/transformer-template-app/MyTransformer.js b/wrappers/s2i/nodejs/test/transformer-template-app/MyTransformer.js
new file mode 100644
index 0000000000..2fae2f4d3b
--- /dev/null
+++ b/wrappers/s2i/nodejs/test/transformer-template-app/MyTransformer.js
@@ -0,0 +1,17 @@
+let MyTransformer = function() {};
+
+MyTransformer.prototype.init = function() {
+  console.log("Initializing Transform ...");
+};
+
+MyTransformer.prototype.transform_input = function(X, names) {
+  console.log("Identity Transform ...");
+  return X;
+};
+
+MyTransformer.prototype.transform_output = function(X, names) {
+  console.log("Identity Transform ...");
+  return X;
+};
+
+module.exports = MyTransformer;
diff --git a/wrappers/s2i/nodejs/test/transformer-template-app/README.md b/wrappers/s2i/nodejs/test/transformer-template-app/README.md
new file mode 100644
index 0000000000..3a86c39a55
--- /dev/null
+++ b/wrappers/s2i/nodejs/test/transformer-template-app/README.md
@@ -0,0 +1,3 @@
+## NodeJs transformer example
+
+This is an example transformer with an identity input and output tansform functions
diff --git a/wrappers/s2i/nodejs/test/transformer-template-app/package.json b/wrappers/s2i/nodejs/test/transformer-template-app/package.json
new file mode 100644
index 0000000000..fa9eb5bad7
--- /dev/null
+++ b/wrappers/s2i/nodejs/test/transformer-template-app/package.json
@@ -0,0 +1,10 @@
+{
+  "name": "transformer-template-app",
+  "version": "1.0.0",
+  "description": "",
+  "main": "MyTransformer.js",
+  "scripts": {
+    "test": "echo \"Error: no test specified\" && exit 1"
+  },
+  "author": "Sachin.Varghese"
+}
diff --git a/wrappers/s2i/nodejs/transformer_microservice.js b/wrappers/s2i/nodejs/transformer_microservice.js
new file mode 100644
index 0000000000..1237da7ea7
--- /dev/null
+++ b/wrappers/s2i/nodejs/transformer_microservice.js
@@ -0,0 +1,133 @@
+const express = require("express");
+const app = express();
+const bodyParser = require("body-parser");
+const grpc = require("grpc");
+const grpc_services = require("./prediction_grpc_pb");
+
+module.exports = (
+  user_model,
+  api,
+  port,
+  rest_data_to_array,
+  array_to_rest_data,
+  array_to_grpc_data
+) => {
+  if (
+    user_model.transform_input &&
+    typeof user_model.transform_input === "function"
+  ) {
+    console.log("Transform function loaded successfully");
+  } else if (
+    user_model.transform_output &&
+    typeof user_model.transform_output === "function"
+  ) {
+    console.log("Transform function loaded successfully");
+  } else {
+    console.log(user_model);
+    process.exit(1);
+  }
+  let transform_input = user_model.transform_input
+    ? user_model.transform_input.bind(user_model)
+    : null;
+  let transform_output = user_model.transform_output
+    ? user_model.transform_output.bind(user_model)
+    : null;
+
+  if (api === "REST") {
+    app.use(bodyParser.urlencoded({ extended: true }));
+    app.post("/transform-input", (req, res) => {
+      try {
+        body = JSON.parse(req.body.json);
+        body = body.data;
+      } catch (msg) {
+        console.log(msg);
+        res.status(500).send("Cannot parse transform input json " + req.body);
+      }
+      if (transform_input && typeof transform_input === "function") {
+        result = transform_input(rest_data_to_array(body), body.names);
+        result = { data: array_to_rest_data(result, body) };
+        res.status(200).send(result);
+      } else {
+        console.log("Transform function not Found");
+        res.status(500).send(null);
+      }
+    });
+    app.post("/transform-output", (req, res) => {
+      try {
+        body = JSON.parse(req.body.json);
+        body = body.data;
+      } catch (msg) {
+        console.log(msg);
+        res.status(500).send("Cannot parse transform input json " + req.body);
+      }
+      if (transform_output && typeof transform_output === "function") {
+        result = transform_output(rest_data_to_array(body), body.names);
+        result = { data: array_to_rest_data(result, body) };
+        res.status(200).send(result);
+      } else {
+        console.log("Transform function not Found");
+        res.status(500).send(null);
+      }
+    });
+    var server = app.listen(port, () => {
+      console.log(`NodeJs REST Microservice listening on port ${port}!`);
+    });
+    function stopServer(code) {
+      server.close();
+      console.log(`About to exit with code: ${code}`);
+    }
+    process.on("SIGINT", stopServer.bind(this));
+    process.on("SIGTERM", stopServer.bind(this));
+  }
+
+  if (api === "GRPC") {
+    function inputEndpoint(call, callback) {
+      let data = call.request.getData();
+      let body = { names: data.getNamesList() };
+
+      if (data.hasTensor()) {
+        data = data.getTensor();
+        body["tensor"] = {
+          shape: data.getShapeList(),
+          values: data.getValuesList()
+        };
+      } else {
+        body["ndarray"] = data.getNdarray();
+      }
+      result = transform_input(rest_data_to_array(body), body.names);
+      callback(null, array_to_grpc_data(result, body));
+    }
+    function outputEndpoint(call, callback) {
+      let data = call.request.getData();
+      let body = { names: data.getNamesList() };
+
+      if (data.hasTensor()) {
+        data = data.getTensor();
+        body["tensor"] = {
+          shape: data.getShapeList(),
+          values: data.getValuesList()
+        };
+      } else {
+        body["ndarray"] = data.getNdarray();
+      }
+      result = transform_output(rest_data_to_array(body), body.names);
+      callback(null, array_to_grpc_data(result, body));
+    }
+    var server = new grpc.Server();
+    server.addService(grpc_services.TransformerService, {
+      transformInput: inputEndpoint
+    });
+    server.addService(grpc_services.OutputTransformerService, {
+      transformOutput: outputEndpoint
+    });
+    server.bind("0.0.0.0:" + port, grpc.ServerCredentials.createInsecure());
+    server.start();
+    console.log(`NodeJs GRPC Microservice listening on port ${port}!`);
+    function stopServer(code) {
+      server.forceShutdown();
+      console.log(`About to exit with code: ${code}`);
+    }
+    process.on("SIGINT", stopServer.bind(this));
+    process.on("SIGTERM", stopServer.bind(this));
+  }
+};
diff --git a/wrappers/s2i/python-ngraph-onnx/Makefile b/wrappers/s2i/python-ngraph-onnx/Makefile
index 4c5265a503..f9fa9abceb 100644
--- a/wrappers/s2i/python-ngraph-onnx/Makefile
+++ b/wrappers/s2i/python-ngraph-onnx/Makefile
@@ -3,22 +3,9 @@ IMAGE_NAME = docker.io/seldonio/seldon-core-s2i-python3-ngraph-onnx
 
 SELDON_CORE_DIR=../../..
 
-
 .PHONY: build
 build: 
 	docker build -t $(IMAGE_NAME):$(IMAGE_VERSION) .
 
 push_to_dockerhub:
 	docker push $(IMAGE_NAME):$(IMAGE_VERSION)
-
-
-.PHONY: test
-test:
-	docker build -t $(IMAGE_NAME)-candidate .
-	IMAGE_NAME=$(IMAGE_NAME)-candidate test/run
-
-.PHONY: clean
-clean:
-	rm -rf test/model-template-app/.git
-	rm -rf test/router-template-app/.git
-	rm -rf test/transformer-template-app/.git
diff --git a/wrappers/s2i/python-tensorrt/Dockerfile b/wrappers/s2i/python-tensorrt/Dockerfile
new file mode 100644
index 0000000000..16c4123cd5
--- /dev/null
+++ b/wrappers/s2i/python-tensorrt/Dockerfile
@@ -0,0 +1,21 @@
+FROM seldonio/seldon-core-s2i-python3:0.2
+
+RUN apt-get update && apt-get install -y \
+    	build-essential libssl1.0-dev libopencv-dev libopencv-core-dev python-pil \
+	software-properties-common autoconf automake libtool pkg-config
+
+WORKDIR /home
+
+RUN git clone --single-branch -b change_ld_flags https://github.com/cliveseldon/dl-inference-server.git
+
+RUN pip install --no-cache-dir --upgrade setuptools grpcio-tools
+
+RUN cd dl-inference-server && \
+    make -j4 -f Makefile.clients all pip
+
+RUN pip install --no-cache-dir --upgrade dl-inference-server/build/dist/dist/inference_server-0.5.0-cp36-cp36m-linux_x86_64.whl
+
+RUN rm -rf dl-inference-server/build
+
+WORKDIR /microservice
+
diff --git a/wrappers/s2i/python-tensorrt/Makefile b/wrappers/s2i/python-tensorrt/Makefile
new file mode 100644
index 0000000000..cb171a045b
--- /dev/null
+++ b/wrappers/s2i/python-tensorrt/Makefile
@@ -0,0 +1,12 @@
+IMAGE_VERSION=0.1
+IMAGE_NAME = docker.io/seldonio/seldon-core-s2i-python3-tensorrt
+
+SELDON_CORE_DIR=../../..
+
+.PHONY: build
+build: 
+	docker build -t $(IMAGE_NAME):$(IMAGE_VERSION) .
+
+push_to_dockerhub:
+	docker push $(IMAGE_NAME):$(IMAGE_VERSION)
+