I am an engineer specializing in Deep Learning, Computer Vision, and AI Agents. I build production-grade applications that solve real-world problems—from real-time market data pipelines to autonomous code review systems.
- 🔭 Currently building: SyntaxAgent (an autonomous GitHub code review agent) and exploring Swift for macOS development.
- 🛠️ Technical Focus: Local LLMs, RAG Systems, Time-Series Forecasting (LSTMs), and React/Next.js.
- 💡 Passionate about bridging the gap between complex ML models and intuitive user interfaces.
- SyntaxAgent: A self-hosted AI code review agent. Orchestrates GitHub PR webhooks via FastAPI/NestJS and generates feedback using vLLM.
- RAD-Knowledge-Assistant: A Retrieval-Augmented Generation (RAG) system built in Python to answer user queries using private document collections with minimal hallucination.
- Ai-Powered-Legal-Summarizer: Leverages transformer-based models to generate concise, accurate summaries of complex legal documents.
- ATS-resume: An NLP-powered tool that extracts information from resumes and generates ATS-optimized versions tailored for specific job positions.
- sign-language-interpreter: Real-time sign language interpretation using MediaPipe for hand gesture detection and a custom neural network.
- NutriBot: A web app that analyzes food packaging images via computer vision to determine health scores and classify diet types (vegan/vegetarian).
- Dog_Breed: An EfficientNetB3 deep learning model with transfer learning to classify 120 different dog breeds, featuring a Streamlit UI.
- Nifty-50-Live-Pulse: Real-time market data pipeline architecture.
- Nifty-500-predictions: A deep learning model using LSTM neural networks analyzing 5 years of historical data to predict the top 10 most profitable stocks.
- TAXBOT: An AI-powered Indian Income Tax Assistant built with Streamlit for tax calculation and planning.
Languages: Python • TypeScript/JavaScript • SQL
AI & ML: TensorFlow • Keras • LLMs (vLLM, Ollama) • MediaPipe • RAG Pipelines
Backend & Data: FastAPI • Node.js • Streamlit • Jupyter • Pandas/NumPy
Frontend: React • Tailwind CSS


