I'm a fifth year Ph.D. candidate in Statistics at Stanford,
advised by Professor John Duchi and supported
by Stanford Graduate Fellowship. I am
very fortunate to be supervised by Professor
Andrea Montanari and Professor Rob Tibshirani
on particular projects.
My main research interest is in designing efficient machine learning algorithms to address three challenges in modern Data Science:
 How to generate statistically efficient procedures? In classical problems, a general recipe for obtaining an efficient estimator is to maximize the likelihood. But this approach cannot be used if our application requires say differential privacy or if our algorithms must satisfy communication constraints.
 How to wisely tradeoff between computational and statistical efficiency? For example, the solution of nonconvex objectives are often statistically efficient. But how can we design nonconvex objectives that are amenable to available optimizers?
 How to recover lowdimensional signals from high dimensional data? Existing techniques address the problem under a parametric setting. Can we solve this problem in a nonparametric setting? What are the limits?
My research aims to answer the above questions. See the research page for details.
Here is my CV.
