Enterprise AI Knowledge Assistant | Hybrid RAG System

Designed and built a production-ready RAG (Retrieval-Augmented Generation) system that turns any document into a chatbot with cited, hallucination-free answers. Combines FAISS semantic search with BM25 keyword retrieval for higher accuracy, supports multi-source ingestion (PDF, DOCX, CSV, URLs), and maintains conversation memory per session. Powered by Groq's LLaMA 3.3 70B for fast inference. Delivered as a Next.js web app, a Telegram bot, and a documented REST API — deployed via Docker on Hugging Face Spaces and Vercel.