Overview
A Retrieval-Augmented Generation (RAG) system built specifically for financial document QA. Point it at an annual report and ask it anything — revenue breakdown, risk factors, segment performance.
Architecture
Two main flows: document ingestion and question answering. Ingestion handles parsing, chunking, embedding, and storing vectors. QA handles retrieval and generation.
Stack
Python, FastAPI, LangChain, Pinecone, OpenAI, sentence-transformers.