A Retrieval-Augmented Generation (RAG) chatbot that answers questions about SEC 10-K filings for major tech companies. The system processes and stores SEC filings (text + XBRL) and provides well-reasoned responses using Llama 3.3 model.

Built a multi-functional OCR tool using Llama 3.2 Vision via OpenRouter API for advanced image understanding and real-time text extraction. Integrated Streamlit for the frontend, with custom workflows to support: Text and LaTeX extraction from images, Code and chart recognition, Visual Q&A and document intelligence

Built a Streamlit platform to identify fentanyl compounds using Python, FastAPI, and PostgreSQL. Scraped and cleaned 11K+ entries from public sources to help researchers find what they need.

Built a Chrome extension to clean and enhance AI prompts using free LLMs that runs locally. One click, better results, no backend.

Created as a starter kit for folks who want to build sleek, fast websites without losing their minds. Built with React, Vite, and Tailwind CSS.