I build systems that live at the intersection of computer vision, ML engineering, and full-stack development — everything from real-time inference on edge hardware to AI-powered web apps.
Computer Vision & Robotics
Low-latency perception systems for robotics using YOLO and C++ on Jetson Orin/Nano hardware. Right now I'm deep in camera calibration, undistortion pipelines, and squeezing real-time performance out of constrained edge devices — the kind of work where milliseconds actually matter.
ML Engineering & Predictive Modeling
End-to-end ML pipelines, from raw data to deployed models. My projects span sports analytics (NBAnomaly, NBAPredict), financial signal processing (MarketMind), and medical AI (AutoCPT).
Full-Stack & Systems
Production web apps with Next.js 15 and TypeScript, REST/WebSocket backends in Flask and Python, and distributed system design — minus the sharding (for now).
| Project | What it does | Stack |
|---|---|---|
| AutoCPT | Real-time AI assistant for medical CPT code extraction from live speech. Won Best Use of AI at HackNYU. | Python · Flask · React · TypeScript |
| Verdia | Metabolic risk tracker that aligns healthy eating incentives for patients and insurers via the Knot API. Built at Hack Princeton '26. | Python |
| MarketMind | ML-driven market analysis and signal generation. | Python |
| NBAnomaly | Anomaly detection across NBA player and team performance data. | Python |
| UGSRP | Camera calibration and undistortion optimizations for computer vision pipelines. | Python · C++ |
