Design a RAG-based chatbot and write the code in Python.

The solution is identical to the one described in question 3. The key components are:

1. Document ingestion – load PDFs (or other sources) and split into manageable chunks.
2. Embedding – convert chunks into vector representations using an OpenAI or HuggingFace model.
3. Vector store – persist vectors in FAISS or Pinecone for fast similarity search.
4. Retrieval – given a user query, retrieve the top‑k relevant chunks.
5. Generation – feed the retrieved context into a large language model to produce a coherent answer.

A minimal implementation:

# (same code as in question 3)

This architecture scales horizontally: you can shard the vector store, use a distributed LLM inference service, and cache frequent queries. The RAG approach improves factuality because the model grounds its responses in the retrieved documents rather than relying solely on its training data.