Depen Morwani

Hi! I am a fourth-year PhD student in the ML Foundations group at Harvard, advised by Boaz Barak and Sham Kakade. My research focuses on understanding and improving optimization algorithms in deep learning.

Previously, I was a Research Intern at Google DeepMind, where I worked with James Martens on empirical and theoretical studies of the mechanisms underlying progressive sharpening. I have also been a Research Intern at Microsoft Research, collaborating with Cyril Zhang and Adam Block to study the interaction between learning rate decay, weight averaging, and sharpening.

Before my PhD, I completed my Master’s degree at IIT Madras under the guidance of Harish Guruprasad, where we investigated the implications of weight normalization. I also spent a year as a Research Associate at Google Research India, working with Praneeth Netrapalli and Prateek Jain on understanding the simplicity bias induced by gradient descent.