I build machine-learning systems that survive production: RAG pipelines, LLM-powered automation, computer vision, and the real-time APIs and AWS infrastructure underneath them.
Equally at home training a model and shipping the product around it: FastAPI & Flask on the backend, React on the front, Qdrant, PostgreSQL and MongoDB underneath.
Real-time quote generation over WebSockets on FastAPI + React, grounded by a Qdrant RAG pipeline with hybrid semantic + BM25 retrieval.
End-to-end computer vision: YOLO detection at 91% accuracy, OCR, and a custom pixel-traversal algorithm that maps equipment-instrument links.
Email-triggered N8N and Zapier automations: Drive files summarized into Docs, and a housing-search bot that scrapes Airbnb and Furnished Finder and replies with the right property.