I built an air-gapped semantic search engine for MCEME that indexes college archives, technical reports, and engineering diagrams using local LLMs and ChromaDB, with no external API dependencies.
At IBM, I develop enterprise agentic frameworks and advanced RAG pipelines, and design and benchmark systems for high-concurrency retrieval and semantic search. At AURA AI, I engineered a hybrid search system using BM25 and Cross-Encoder re-ranking, and integrated TrackNet-based training and tracking pipelines into a sports video analytics workflow.
At Kenko AI, I optimized cross-platform CMS systems and internal API integrations for a healthcare tech content pipeline. My work also includes a digital-twin policy simulation with real-time knowledge graphs and an autonomous agent platform with human-in-the-loop approvals.

