Home

Assistant Professor
Department of Statistics
University of Connecticut

My research combines statistics and machine learning to develop robust, scalable systems for spatiotemporal processes. This work is primairly driven by problems in climate science.

Research Areas

Climate Model Evaluation and Integration
My research develops statistical and machine learning methods for evaluating climate models against observations, with a focus on comparing long-run climate distributions and constraining future projections. As a counterpart to this work, I also work on deep learning / neural operator approaches that integrate physical model output with observational data to constrain climate projections and reduce bias.

Uncertainty Quantification and Generative Modeling
I am also working on conformal and distribution-free uncertainty quantification methods for neural operators more generally, with recent efforts connecting conformal inference to generative models that represent full predictive distributions. Ongoing research explores calibrated flow-based generative models as efficient alternatives to autoregressive climate emulators.

Structure Discovery in Climate Systems
Finally, my group is also develops statistical and machine learning methods to uncover dependence structures, causal relationships, and compound risk in large-scale climate systems, enabling more reliable characterization of extremes and rare events.