CS + Data Science at NYU, class of '27, with a cybersecurity minor. I like hard problems, shipping fast, and people who do both.
Now: researching attack surfaces in agentic and decentralised systems at NYU's OSIRIS Lab. Last summer I was at Clear Street doing backend and application security work: a Go log pipeline on Kinesis and an agent harness that reads code property graphs to find vulnerabilities across a monorepo. Before that, first engineering intern on the forward deployed team at Forkast (Antler '25).
Things I've built
- Rewind — time-travel debugger for coding-agent sessions. Scrub back to any step, see exactly what the model saw, fork with a different model, diff the branches. Live here.
- Orchard — distributed inference and federated fine-tuning across the Apple Silicon machines you already own. Top 10% of YC applicants.
- Ouroboros — runtime detection for MCP tool calls. Catches tool poisoning, cross-server shadowing, and rug-pulls before they run.
- What The Rent?! — Chrome extensions for StreetEasy and Zillow that surface the hidden costs of a rental in NYC and SF. Won Microsoft x Musa Capital and Khosla Ventures x ForgeHacks. On the Chrome Web Store.
- Trevor AI — call a phone number, sub-agents research and trade for you while you talk.
- Plop — any room into a 2D/3D interior design with Qwen-VL and real, swappable products. Built in three hours, won the YC Startup School hackathon.
- Yieldly — farms upload what they've got, farm-to-table restaurants get AI-priced menus from it. Won the Antler hackathon.
- NutriMap — an agent over scraped NYC menus that picks your meals from your macros. People's Choice at Google x HackNYU.
Before that: Exar North (agentic finance research platform, 300+ write-ups a week), GoTrust (GraphRAG NL-to-SQL on a 3B model, on-prem), Kanlet (full-stack CRM feature), Ambee (time-series ILI forecasting), UPL (threat intel, traced a live incident). Research with Harvard Business School and NYU Langone.
Off-screen: photography, gym, guitar.