Projects

Upload any doc — PDF, DOCX, or TXT — and chat with it. Built a full RAG pipeline with FAISS vector search, MMR retrieval, and Groq's LLaMA to give accurate, source-grounded answers from your own files.

Track expenses through natural language — no dashboards, no UI. Built an MCP server that plugs straight into Claude Desktop, Cursor, or Gemini CLI. Just talk to your AI and it logs, edits, or summarizes your spending. Fully local, SQLite-backed, zero cloud.

Pick a movie, get five more you'll actually want to watch. Built a content-based recommendation engine using TF-IDF and cosine similarity, with live posters and details pulled from the TMDb API.

Upload a CSV or Excel file and just ask it questions. Built an AI-powered data analysis tool where you talk to your dataset in plain English and get back real insights — trends, stats, correlations — powered by Groq's LLaMA 3.