Autonomous Multi-Agent Researcher

Architected a sequential state-machine pipeline with three specialized LLM agents: a Planner, a Researcher, and a Synthesizer to compile fully cited Markdown intelligence reports.

Deliberately separates planning, retrieval, and synthesis across distinct agents to prevent context degradation and confabulation.

Implements rate-limit safeguards and graceful degradation logic so the pipeline continues producing partial results.

Hosted on Hugging Face Spaces as a Streamlit application with persistent containerized deployment.