RAG as a Service (RAGaaS)

A Flask-based API service that provides Retrieval-Augmented Generation (RAG) capabilities as a service. This project allows you to upload documents, create knowledge bases, and query them using natural language, leveraging the power of large language models and vector databases.

Features

• 📚 Document Upload: Support for text and CSV file uploads via URLs
• 🔍 Vector Search: Utilizes PGVector for efficient similarity search
• 🤖 Multiple LLM Support: Compatible with OpenAI models and custom endpoints
• 🗄️ PostgreSQL Integration: Persistent storage for embeddings and chat history
• 📝 Chat History: Maintains conversation context and history
• 📊 Swagger Documentation: API documentation and testing interface

Tech Stack

• Backend Framework: Flask
• Database: PostgreSQL with pgvector extension
• Vector Embeddings: OpenAI Embeddings (text-embedding-3-large)
• LLM Integration: LangChain
• API Documentation: Flasgger/Swagger
• Containerization: Docker

Prerequisites

• Docker and Docker Compose
• OpenAI API Key or compatible endpoint
• PostgreSQL with pgvector extension

Environment Variables

Copy .env.example to .env and configure the following variables:

OPENAI_API_KEY=your_api_key
VECTOR_DB_USERNAME=postgres
VECTOR_DB_PASSWORD=postgres
VECTOR_DB_HOSTNAME=postgres
VECTOR_DB_DATABASE=ragaas
VECTOR_DB_PORT=5432

Installation & Setup

1. Clone the repository
2. Copy .env.example to .env and configure your environment variables
3. Build and run the services:

docker-compose up --build

The API will be available at http://localhost:8443

API Documentation

The API documentation is available through Swagger UI at:

• http://localhost:8443/apidocs

You can use this interactive interface to:

• Explore all available endpoints
• Test API endpoints directly
• View request/response schemas
• Download OpenAPI specification

API Endpoints

• GET /: Welcome endpoint with service status
• POST /upload-file: Upload documents to create knowledge base
  • Required parameters:
    • collection_id: Unique identifier for the document collection
    • file_url: URL of the file to process (supports .txt and .csv)
• POST /chat-message: Chat endpoint for querying the knowledge base
  • Supports conversation history and context

Usage Example

1. Upload a document:

curl -X POST http://localhost:8443/upload-file \
  -H "Content-Type: application/json" \
  -d '{"collection_id": "my_docs", "file_url": "https://example.com/document.txt"}'

2. Query the knowledge base:

curl -X POST http://localhost:8443/chat-message \
  -H "Content-Type: application/json" \
  -d '{"collection_id": "my_docs", "user_input": "What does the document say about X?"}'

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.