Sanjeev Raja

I’m a first-year Computer Science PhD student at Berkeley AI Research (BAIR) advised by Aditi Krishnapriyan.

My interests lie at the intersection of machine learning and the natural sciences. On the applied side, I am interested in combining the expressivity of modern deep neural networks with the generalization capabilities of first-principles scientific models to improve partial differential equation solvers, molecular and fluid dynamics simulations, and other scientific computing workflows. On the fundamental side, I am interested in understanding how scientific inductive biases (e.g conservation or equivariance constraints) impact the generalization capabilities of machine learning models in both data-limited and data-rich settings.

Most recently I was a research intern at Lawrence Berkeley National Laboratory, where I worked with Anima Anandkumar and Kamyar Azizzadenesheli on memory-efficient multigrid neural operators for weather and fluid flow prediction. I also represented my university in ProjectX, an international research competition on ML for climate change solutions, for which we developed adversarial super-resolution methods for global climate methods.