I build practical AI/ML systems across RAG, computer vision, and medical imaging — and turn models into usable software: APIs, evaluation pipelines, Dockerized services, and clean interfaces.
A few things I've built recently.
Jul 2026
0 fabricated facts · 100% abstention on unanswerable questions
Production-style RAG system for domain PDFs: FastAPI, ChromaDB, and Claude with schema-constrained generation. Answers are grounded in per-claim citations and the system abstains when the documents don't contain the answer.
Jun 2026
94% accuracy / 0.94 macro-F1 on reproducible benchmark
Production-style PyTorch image classification service: shared train/serve preprocessing, FastAPI inference with Grad-CAM explanations, and honest evaluation on a reproducible synthetic benchmark (non-clinical).
Jan – May 2025
Two-sided marketplace, web + mobile
Peer-to-peer parking marketplace: homeowners rent driveways; drivers book affordable spots.
From models and evaluation through APIs, Docker, and CI.
I'm open to ML/AI engineering and data science roles, collaborations, and interesting projects. The fastest way to reach me is email.