I’m a PhD student in the Mathematical Foundations of Deep Learning Lab, advised by Prof. Chaoyue Liu. My research focuses on aspects of optimization theory, intrinsic interpretability, & domain generalization within deep learning.

I like to approach research through rigorous mathematical foundations. Currently, I’m exploring the interplay of interpretability and robustness, and am developing methods that satisfy theoretically-motivated correctness criteria.

These guarantees enable the effective use of interpretability techniques as tools for model optimization. Typical approaches that regularize for ground-truth explanations malicious compliance – in that matching ground-truth explanations is possible without being truly robust to the spurious factors that the regularizer is intended to suppress. For explanation techniques with correctness guarantees, this failure mode is entirely absent.

Besides deep learning, I also enjoy hiking, motor racing, chess, swimming, mountain biking, surfing (unfortunately I currently suck at those last two); really anything that gets the adrenaline pumping.

If you’re doing research along similar lines, or just generally find this stuff interesting, I’d love to chat and exchange perspective. I’m active on email and matrix!