My research is focused on unsupervised learning, sublinear time algorithms and information theory. My work has led to publications at top CS theory conferences like STOC, FOCS and SODA.

I built survival models to predict likelihood of wind turbine failure, with the goals being predictive maintenance, resource allocation and inventory management. The false positive rate was reduced by 70% compared to previous models. The models were built using turbine features and multivariate time series data.

Derived some spectral properties of expander graphs by analyzing interlacing families of polynomials, with Navin Goyal.

Constructed a polynomial-time algorithm to approximate Kostka numbers up to a constant factor, with Markus Bläser. Awarded DAAD-WISE Fellowship by the German Academic Exchange Service (awarded to 150 Indian students).