astronomer/ask-astro

Implements Retrieval Augmented Generation (RAG) with Airflow DAGs that ingest data from eight sources into Weaviate, then uses LangChain's ConversationalRetrievalChain with prompt rewording, document reranking, and relevancy filtering to generate responses. Includes automated feedback loops where user ratings and LLM-generated quality scores continuously improve the knowledge base, plus a Slack bot and public UI for access.

280 stars. No commits in the last 6 months.