I am interested in learning and generalization from data from a diverse set of perspectives: computational, inferential, decisional, and mathematical. Learning and generalization from data helps addressing scientific questions from ecology to neuroscience, and tackling challenges to build the next generation of AI technology. 

I have recently been working on learning under distributional shift, and more broadly on learning maps to transform data.  I have also been working on developing practical tools to address the trade-off challenges arising from upscaling or downscaling generative models in various AI domains including language and vision.