I am currently a Member of Technical Staff at OpenAI.

From 2021-2024, I was a post-doctoral Miller Fellow at UC Berkeley working in the Department of Statistics. I was hosted by Martin Wainwright and Ryan Tibshirani, and I am grateful to have been mentored by and closely collaborated with Yun Song and Song Mei. In 2021, I obtained my PhD in Statistics from Stanford University, advised by Andrea Montanari and supported in part by the NSF Graduate Research Fellowship. My dissertation won the Theodore W. Anderson Theory of Statistics Award. I received a bachelor’s degree in mathematics and physics in 2014 and a master’s degree in electrical engineering in 2016, both from Stanford University.

In the summer of 2019, I held an internship at Microsoft Research New England, and was fortunate to be hosted by Vasilis Syrgkanis and Greg Lewis. In Fall 2021, I participated as a visiting postdoc in the Computational Complexity of Statistical Inference program at the Simons Insitute. From April 2020 to September 2021, I helped organize the Online Causal Inference Seminar. Outside of statistics, I enjoy biking, cooking, and reading.

Research interests

I develop algorithms for analyzing high-dimensional data. I am particularly interested in how errors from complex and high-dimensional prediction models lead to bias and overconfidence in downstream analyses. I developed methods to correct this bias and correctly calibrate inferences. I also work on average-case analysis of first-order methods on non-convex objectives (here, here) and variational Bayesian inference (here, here, here). My work intersects with semiparametrics, high-dimensional causal inference, exact asymptotics, debiased machine learning, and Gaussian processes.

In the short term, my primary research objective is to flesh out a rich theory and methodological toolbox for semi-parametric estimation and inference in ‘‘inconsistency regimes.’’ These are regimes in which nuisance parameter are high-dimensional and unstructured, so cannot be estimated well: see here, here, here.

Since Spring 2022, I have worked with Yun Song on tree-based methods to estimate fitness in evolving populations, and am broadly interested in high-dimensional statistical problems in biology.