Elynn Chen
Assistant Professor of Technology, Operations and Statistics, NYU Stern School of Business
Modern business runs on messy, high-dimensional, ever-shifting data — and the decisions built on that data are only as good as the statistics underneath them. My research builds those foundations: methods that preserve structure, adapt to heterogeneity, and quantify uncertainty, so that companies and AI systems can make reliable decisions rather than confident guesses.
I am an Assistant Professor of Technology, Operations and Statistics (TOPS) at NYU Stern School of Business. My work spans three connected streams: structured multi-way learning for matrix, tensor, and network data; learning and decision-making under heterogeneity — reinforcement learning, bandits, and dynamic pricing across diverse populations and environments; and trustworthy AI for business, from revenue management to generative and agentic AI. This work appears in JASA, JRSS-B, the Annals of Statistics, and Management Science, and at NeurIPS, ICML, ICLR, and AISTATS. It is supported by an NSF DMS Research Award (2024–2027), following an earlier NSF Postdoctoral Fellowship. I serve as an Associate Editor of ACM TKDD and an Area Chair for NeurIPS.
My path here was a winding one: a Ph.D. in Statistics at Rutgers with Rong Chen, a postdoc with Jianqing Fan at Princeton ORFE, time as a research fellow at OpenAI working on latent heterogeneous reinforcement learning for societal applications, and a postdoc with Michael I. Jordan in EECS at UC Berkeley. That mix of theory and practice shapes how I pick problems: rigorous enough to prove something, real enough to matter.
The part of the job I’m proudest of is mentoring. Much of my work is co-authored with students I advise, who have gone on to faculty positions (City University of Hong Kong) and doctoral programs at Stanford GSB, Dartmouth, and Purdue. I’m actively recruiting motivated students — see the Students page for openings and how to apply.
For the most up-to-date list of publications, see my Google Scholar profile.
news
courses, tutorials and books
- Reinforcement Learning Notes — from MDP foundations to RL for LLM.
recent research posts
- Same Logic, Different Words: Why AI Systems Contradict Themselves, and What We Did About It — Large language models often change their answer when a rule is reworded without changing its meaning. In a ... (Nov 2026)
- Optimize Once, Execute Fast: Why AI Teams Make You Wait, and How to Fix the Org Chart — AI systems built from several cooperating models give better answers but take longer to deliver them. In a ... (Nov 2026)