Stateful Evidence-Driven Retrieval-Augmented Generation with Iterative Reasoning

Build a modular RAG agent that transforms retrieved chunks into 'reasoning units' before generation. Focus on an extensible interface for different logic extractors.

Suggested repo: reason-rag

"Stop feeding raw text to your LLM; feed it structured reasoning steps."

Estimated effort: 60h