A production-oriented RAG pipeline combining dense vector search and keyword search fused with Reciprocal Rank Fusion, cross-encoder reranking, and grounded answer generation with enforced citations. Built on PostgreSQL with pgvector and Redis, deployed on AWS with Docker and FastAPI. Includes an evaluation harness measuring hit rate, MRR, faithfulness, and answer relevance so every tuning decision is defensible, not vibes-based.