FareedKhan-dev/temporal-ai-agent-pipeline

Implements a six-stage temporal knowledge graph pipeline using LangGraph workflows: semantic chunking, atomic fact extraction with timestamps, entity resolution for deduplication, temporal invalidation to mark outdated facts, and graph construction—enabling RAG systems to query time-aware knowledge. Built on LangChain with OpenAI LLMs and SQLite storage, it processes evolving datasets (financial transcripts, medical codes) to prevent hallucinations in dynamically updated knowledge bases.

No commits in the last 6 months.