Hi, I'm Shyamal, an IFML postdoctoral fellow at UT Austin. Previously, I was fortunate to be advised by Cliff Stein and Xi Chen in the Theory of Computation Group at Columbia University. My research interests are in computational learning theory, property testing, and sublinear algorithms.

I'm broadly interested in questions such as:

1. Given labeled data, how much time do algorithms need to learn a high-accuracy hypothesis that predicts the labels of (unlabelled) examples?

2. Given query access to a function, how many queries do we need to determine if it's structured?

3. What metrics are well-approximated by the shortest paths of a sparse graph?

Some specific problems I have thought about are learning intersections of halfspaces, learning DNF, tolerant junta testing, distribution-free testing of boolean functions, and spanners.