Design a Retrieval-Augmented Generation (RAG) system for production use. The system should efficiently index large document corpora into a vector store, retrieve semantically relevant context at query time, and augment LLM responses with that context to improve accuracy and reduce hallucinations.