RashedDoha/langchain_rag_demo

A basic retrieval augmented generation pipeline (RAG) with a FastAPI streaming output endpoint.

It does the following functions

• Ingests data in the form of documents from the data/raw directory
• Chunks the documents with appropriate overlap for improved retrieval accuracy
• Stores the chunks in a vector store (Chromadb) to accomplish similarity search on a given query
• Retrieves information from the vector store given a query string
• Invokes an LLM to generate answer to the given user question by augmenting context from the retrieval process