I built an end-to-end hybrid RAG (Retrieval-Augmented Generation) chatbot that answers questions over custom document collections. The system combines local semantic embeddings with LM Studio embeddings for accurate retrieval, uses FAISS for fast vector search, and generates responses with a large language model. It includes a FastAPI backend, document upload support (PDF/TXT), and a modern React chat interface similar to ChatGPT, with conversation history and Markdown rendering.