AI infrastructure · Backend systems · Open-source engineering

I build the infrastructure that makes AI systems reliable: data pipelines, inference services, persistent state, and the APIs around them. I work from a reproducible problem to a tested implementation, then stay involved through review and integration.

LinkedIn · Email · Open-source contributions

• ComfyUI: persistent execution history — retain completed workflow records across restarts, with retention controls, transactional deletion, and privacy-aware storage. Open PR.
• Chroma: native sparse search — add exact sparse search to the embedded query path. Draft PR.
• LanceDB: streaming checkpoint provenance — make checkpoint identity and restore compatibility explicit. Draft PR.

• zero-hour — supply-chain incident triage for coding agents, connecting security advisories to dependency and file-level evidence.
• Japanese OCR for RAG — extract Japanese PDFs into traceable text for retrieval pipelines, processing large documents page by page.

My work spans machine learning, backend services, and cloud infrastructure: forecasting, recommendation systems, document intelligence, APIs, and deployment automation. I focus on correctness under failure, measurable performance, and compatibility with the systems people already use.

Tools: Python, TypeScript/JavaScript, Go, FastAPI, PostgreSQL, AWS, Docker, and GitHub Actions.

M.S. in Computer Science, New York Institute of Technology · B.Tech. in Computer Engineering, UKA Tarsadia University.

Building AI infrastructure or working through a difficult systems problem? Let's talk.