I wish to understand the foundations of learning and decision making towards developing intelligent systems. My approach forges connections between different disciplines, and is often focused on discovering scientifically interpretable structure in data. I am particularly engaged in building methods for probabilistic deep learning, scalable Gaussian processes, physics-inspired machine learning, AI alignment, kernel learning, and training of deep neural networks. I have applied my work to time series, vision, spatial statistics, NLP, counterfactual inference, public policy, medicine, and physics. I also believe in open and reproducible research, and have introduced several software libraries.