About
Career
Global Director, Data & Analytics
Chief Data Scientist for Global Analytics, responsible for the science and engineering behind everything from forecasting to agentic dev tools.
Principal Machine Learning Engineer
Designed the framework that trains hierarchical models at global scale, used by products across the company, from forecasting to marketing.
Senior Engineering Manager
Led the team that built in-house marketing mix models as custom differentiable programs in TensorFlow.
Senior Machine Learning Research Scientist, Ecommerce
Invented similarity regularization, granted as a US patent, which makes similar products behave alike across pricing and marketing models. Patent US12333399B2
Data Scientist, Ecommerce
Built a marketing attribution model on a generalized additive neural network with economic priors, and product embeddings adapted from DeepWalk.
Technical Research Assistant, RiskEcon Lab
Researched machine vision for gait and movement quality, and wrote a survey on gait classification. Paper on arXiv
Master of Science in Computer Science
Wrote a thesis on classifying the quality of movement from motion capture, which Courant chose for publication.
Bachelor of Arts in Mathematics & Computer Science
Alongside it, built a classifier for fashion products at StyleSage and analyzed fMRI data for a poster on brain entropy.
Introduction
Leading
Being a builder is central to how I lead. I still write code, design models, and spend time in the details. Knowing the work deeply helps me see where to take it, make difficult calls, and earn the confidence of the people doing it with me.
Teaching is a big part of that, too. I love the moment when something clicks for someone and they can do something they couldn’t before. I want to help people develop the understanding and judgment to find their own way through hard problems. That includes having good reasons to disagree with me.
Building
I tend to get absorbed in the things I build. I want to pick apart the problem, live in the data, and learn from the people who know the business. That understanding shapes everything from how I represent a problem in a model to what I decide is worth building.
I usually start with a question: what will someone do differently if this works? The answer helps me work backward to what the system needs to do and how we should measure it. Making that connection can be surprisingly hard. But when the measure stays close to the decision, improving the system is more likely to improve something people care about.