About Me

I am a Director of Data Science at Capital One working on projects for large-scale customer graphs for entity resolution, GenAI methods and I'm very passionate about building models that actually make it into production. My background is in computational physics, which is how I originally got into machine learning. These days I’m interested in foundation models for their in-context capabilities (e.g. tabPFN), interpretable architectures for decision making, and ways to use scalable modern ML methods. I lead a fantastic team of data scientists who work across business areas like fraud, servicing, and marketing, and a lot of my daily work involves taking good research ideas and figuring out how to make them useful inside a real enterprise system, which usually entails a lot governance and risk management. On this site I keep notes, side projects, and ideas I’m exploring. Most of this isn’t polished, and that’s intentional—it’s a place for experiments, learning, and ongoing work. If you find anything useful here (or want to chat about ML,science & software engineering), feel free to reach out. I previously received my Ph.D. from Boston University doing research with Pankaj Mehta. You can check my prior recent and older publications and some older ML repos I built.

Recent projects

Self-consistent scalable clustering

A self-consistent clustering approach that learns hierarchical semantics of clusters.

Machine learning review

An introduction to machine learning in the language of physicists. Covers many of the core ideas of ML.

Reinforcement learning quantum control

Using Q-learning, we studied the rich phase diagram of quantum state preparation and it's implications.