I build systems that learn from scale — and lately I've been thinking a lot about what happens after they learn.

Yale PhD (Computer Engineering)
        ↓
Meta (Instagram / FB Search, ranking @ scale)
        ↓
Citadel Securities (ML Research)

For the past several years I've worked on large-scale recommendation and ranking systems — transformer-based foundation models, deep learning ranking, and the embedding infrastructure that makes all of it run in production at Meta. It's the kind of work where the math is elegant and the pager alerts are not.

Before that: a PhD at Yale, with earlier stops at Waterloo and USTC. Publications span EDA (ICCAD, ASP-DAC) and ML (KDD), and an RL-focused internship at Microsoft that pulled me toward sequential decision-making problems.

I'm moving from Meta into quantitative ML research at Citadel Securities. It's a shift from consumer-scale ranking to markets — different feedback loops, different notion of "ground truth," same underlying question of how to build models that make good decisions under uncertainty.

• 📍 New York
• 🔭 Ranking → ML Research
• 💬 Ask me about embedding infra, ranking systems, or why I keep switching fields

This README is a snapshot, not a mission statement — check back, it'll drift as I do. (And yes, Potato is real.)