Skip to content
View BhavikShangari's full-sized avatar

Block or report BhavikShangari

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
BhavikShangari/README.md

πŸ’» Bhavik Shangari - ML & AI Enthusiast

🌟 Welcome to my GitHub profile!
I'm Bhavik Shangari, a passionate Machine Learning and Artificial Intelligence enthusiast with a deep interest in robotics and solving real-world problems using advanced computational techniques. My expertise lies in developing and fine-tuning AI models, with a recent focus on Large Language Models (LLMs) and Generative AI.


πŸ“– About Me

  • πŸŽ“ BTech in Data Science & Artificial Intelligence
    *Indian Institute of Technology (IIT) Bhilai
  • πŸ’‘ Passion: Building solutions that bridge the gap between cutting-edge AI and practical applications.
  • πŸ”¬ Current Focus: Training and deploying multimodal models, with special attention to low-powered devices and robotics.

πŸ† Key Projects

πŸš€ Jetson-VLM 2

  • Trained a multimodal model combining SigLip and DINO V2 vision encoders with LLama 3.2 (1B).
  • Focused on deploying Vision Language Models on low-powered devices like Jetson Nano.
  • Implemented image alignment using Llava 595K v1.5 Data Mixture.
  • Designed and built a robotic arm using computer vision, microcontrollers, and linear algebra.
  • Mimicked human hand movements for imitation learning and assistance for physically-aided individuals.
  • Built a novel Date&Time2Vec embedding method for capturing temporal relations in stock market data.
  • Implemented a Transformer architecture from scratch in PyTorch.
  • Processed 15 years of stock data with 5M+ data points, combining sentiment analysis and transformer architectures.

πŸ₯ CloudPhysician

  • Developed a pipeline for extracting vital signs (e.g., SPO2, ECG) from patient monitor images using YOLO, OCR, and image processing.

πŸ”§ Skills & Expertise

Languages

  • Python, C, C++

Frameworks & Libraries

  • PyTorch, TensorFlow, Hugging Face, Scikit-learn, Keras, OpenCV, NLTK

Domains

  • Deep Learning: CNNs, RNNs, Transformers, GANs, YOLO
  • Generative AI: Fine-tuning LLMs, multimodal integration
  • Computer Vision: Image segmentation, restoration, enhancement
  • Natural Language Processing: Multimodal alignment, retrieval

🎯 What I’m Working On

  1. Training Jetson-VLM for improved performance on vision-language tasks.
  2. Implementing multimodal robotic action control with OpenVLA approaches.

🌐 Connect with Me


πŸ’‘ Let's collaborate on exciting AI and robotics projects! Feel free to explore my repositories.

Pinned Loading

  1. DS250_Project DS250_Project Public

    Stock Market Forecasting using Transformers

    Python 2 1

  2. Gesture-Controlled-Robotic-Arm Gesture-Controlled-Robotic-Arm Public

    We have made a robotic arm that simply works on hand gestures, as you will simply move your hands, the robotic arm will mimic the exact movement and facilitate to pick and place an object

    Python 1 1

  3. Cloudphysician Cloudphysician Public

    Forked from ASK-03/Cloudphysician

    A project aimed to address challenges in ICU care by leveraging machine learning and computer vision. The primary goal is to develop a system capable of extracting vital signs information from pati…

    Python 1

  4. TRISEASE TRISEASE Public

    Forked from DJIITBH/TRISEASE

    Jupyter Notebook 1