Prospective Fall 2027 PhD Students

I expect to take at most one new PhD student this admissions cycle, and may not take any.

My group has historically worked primarily on geometric and equivariant machine learning, particularly for physical and scientific systems. This remains an area of considerable expertise within the group, and I continue to have active projects and collaborations in this space. However, I am not currently looking specifically to grow the group in equivariant or atomistic machine learning.

The questions I am most interested in exploring right now are broader ones about representation and abstraction: What makes a representation useful for reasoning? What should be made explicit or left implicit? How should representations relate information across different scales or levels of description? What information is gained or lost through abstraction, and when does that matter? How do the representations we choose shape the questions we are able to ask and answer?

I am deliberately open about the domain, methodology, and disciplinary background in which these questions arise. You do not need to work in scientific machine learning—or in an area my group has previously worked in—to be a good fit. In fact, I would be particularly interested in hearing from students who encounter related questions from a substantially different perspective or application area.

I don’t yet know where all of these questions will lead, and that is partly the point. For the small number of students I might consider taking this year, I care less about alignment with a particular existing research program than about a shared intellectual interest in these questions and the possibility of discovering something interesting together.

If you are interested in equivariant or geometric machine learning

If your primary goal is to work specifically on equivariant neural networks, geometric deep learning, or their applications to atomistic systems, there are many excellent researchers for whom these remain central areas of current research. Researchers such as Prof. Melanie Weber and Prof. Robin Walters, are two examples, and there are many others working on these methods in materials, molecular modeling, and related areas.

I continue to advise ongoing work in these areas, but they are not the areas in which I am particularly looking to expand my group this admissions cycle.