Kushal is an undergraduate student studying Computer Science and Math at Stanford University. His research interests lie in improving our mechanistic understanding of deep neural networks and answering the question of how preferences and desires can adaptively be communicated to AI systems. In this vein, his research spans learning and generalization, reward modeling, and ‘empirical-theory’ interpretability across various deep generative models. In his spare time, Kushal enjoys doing photography, outdoor adventures, and thinking about how anything from plate tectonics to moral realism works.