A Retrieval-Augmented Generation (RAG) system built on Amazon Bedrock. Admins upload PDFs, which are chunked, vectorized by a Titan Embedding Model, and stored in FAISS. User queries trigger a similarity search to retrieve relevant context for LLM-generated answers.