I'm looking for PhDs, Postdocs, and RAs! See here for more information.

About me

I'm an incoming Assistant Professor in the College of Computing and Data Science (CCDS) at Nanyang Technological University (NTU). Currently I'm completing a Florence Nightingale Bicentenary Research Fellowship in the Department of Statistics at the University of Oxford, where I also completed a postdoc and DPhil previously. I also co-founded a company called Quro Medical, a medical technology company that performs remote patient monitoring in Southern Africa

Contact information

• Email
• Google Scholar
• Twitter
• Github

Research Interests

I'm interested in AI safety, and in building robust and reliable machine learning systems. Some topics I have worked on include:

• Generative models
• Geometric deep learning
• Causal inference
• Uncertainty quantification

I'm also interested in developing better tools for describing and reasoning about AI systems themselves. At present, the mathematical language we use in AI research is often far removed from the actual code we end up implementing. This complicates the development process, limits automation, and makes it difficult to reason about correctness. I'm interested in using tools from applied category theory, programming languages, logic, and proof assistants (e.g. Lean) improve on this. For example, in this talk, I describe a line of work that uses categorical probability to obtain novel methodology for parameterising stochastically group equivariant neural networks.

If any of our interests overlap, I'd love to hear from you!

Background

I'm currently a Florence Nightingale Bicentenary Research Fellow in the Department of Statistics at the University of Oxford. I previously completed a postdoc with Chris Holmes and Arnaud Doucet working on conformal inference and causal machine learning. Before that I completed my DPhil as part of the AIMS CDT under the supervision of Arnaud Doucet and George Deligiannidis. My thesis covered several topics in (primarily) Bayesian machine learning, including Monte Carlo methods and deep generative modelling. Before they left Oxford, I also worked with Frank Wood and Hongseok Yang on topics in probabilistic programming.