React RAG Chatbot Backend Setup

My React front-end already handles the UI; what I need now is the brain behind it. I’m looking for a Retrieval-Augmented Generation chatbot that feels genuinely conversational—able to understand free-form language, keep track of context across multiple turns, and reply intelligently at low latency.

I’d like the entire back-end written in Python. You can choose Django for its batteries-included approach or Flask if you prefer something lighter, as long as the final solution is clean, well-documented, and easy to deploy. The service should expose a secure REST (or WebSocket, if that offers better streaming performance) endpoint that my React app can hit directly.

To keep expectations crystal-clear, here’s what I consider a successful hand-off:
• A RAG pipeline wired to my vector or document store - Pinecone, returning concise, source-grounded answers.
• Natural language understanding, context-aware responses, and multi-turn conversation handling demonstrably working end-to-end.
• React integration snippet (hooks or service file) showing how to call the new API and display streaming answers in the chat window.
• To be hosted on VPS.
• A brief README covering setup, environment variables, and example queries.

If you’ve built fast, production-ready chatbots before—especially with Python and React—let’s talk.