• I currently work as a Postdoctoral Scholar with Dr. Maike Sonnewald in the Computational Climate and Ocean Group at UC Davis. We have submitted a paper exploring the use of explainable artificial intelligence technqiues to gain insight into the physics learned by machine learning models in the context of ocean dynamics, with specific applications to Atlantic Meridional Overturning Circulation (AMOC). My current work involves building a physically interpretable ocean emulator.

  • In my dissertation, advised by Dr. Ari Stern, and titled ‘Functional Equivariance and Backward Error Analysis’, I present the modified vector field perspective of functional equivariance, develop an algebraic characterization of functional equivariance and formalize conjugate functionally equivariant integrators.

  • I worked with Dr. Donsub Rim to develop a novel low rank expansion of neural networks. We postulate that instabilities in neural networks, in the form of adversarial examples, can be generated and studied using this mathematical expansion. We applied this expansion to a convolutional neural network (CNN) for tsunami early warning and found adversarial perturbations comparable to those obtained using projected gradient descent and developed filters for noisy input.