Masters in Computer Science at NYU. I build things people step inside — spatial computing on Vision Pro, retrieval systems that actually answer, computer vision that reads a heartbeat off a webcam. None of it is about music. All of it was built with a record on.
Open to full-time software / ML engineering roles. Will relocate for the right lineup.
A browser-based collaborative IDE: multi-user real-time editing on Yjs CRDTs over WebSocket with live cursor presence, Monaco with full IntelliSense, an xterm.js + node-pty terminal — and an Orchestrator AI layer (file locking, MD5 conflict detection, 3-tier agent flow) keeping concurrent agents out of each other's way. Dockerised, deployed to Fly.io.
A documentation engine for Robot Operating System: 8+ GitHub repos and 50+ articles pulled through a Dockerised ETL pipeline, Llama 3.2-1B fine-tuned with LoRA on 4,500+ Q&A pairs, Qdrant vector search over 15,000+ chunks.
Reads your pulse off a video feed. GREEN-channel rPPG with FFT signal analysis and 512-frame buffering — 60% less compute than multi-channel ICA — in a native SwiftUI macOS app tracking faces live at 30fps across 56+ test subjects.
I got into engineering the same way I got into records — by taking something apart to find out why it works. A good album and a good system have the same shape: a strong opening, layers you only notice on the third pass, and nothing in there that isn't earning its place.
Right now I'm building VR training tools for chip design at VIDA, fresh off a Masters in CS at NYU. Before that: home automation in Mumbai, VR leasing in New York. I care most about interfaces that make hard things feel physical — a heart rate you can see, a fabrication process you can walk through, documentation that answers back.
Crate-digging, live shows around the city, and an ongoing argument about whether the mastering ruined the reissue. Ask me about it — it's the fastest way to get a long answer out of me.