Built an AI-powered research system that automates end-to-end workflows including literature discovery, evidence validation, and structured report generation using multi-agent pipelines.
- Designed modular architecture combining CLI tool and lightweight API with real-time progress streaming (SSE).
- Implemented multi-agent orchestration with specialized sub-agents for literature retrieval, reasoning validation, and citation integrity.
- Developed hybrid retrieval pipeline integrating ArXiv, Semantic Scholar, and PubMed with rule-based validation.
- Engineered scalable execution with parallel processing, disk-backed state management, and dynamic model routing across LLM providers.