I grew up and went to school in Sussex, during which time I represented Britain at the International Physics Olympiad. I then studied Natural Sciences at Trinity College, University of Cambridge, where I specialised in Physics. I obtained my DPhil in Astrophysics at the University of Oxford, where I was the Graduate Teaching and Research Scholar in Physics at Oriel College. I moved to Paris in 2022 for a postdoctoral position in cosmology and artificial intelligence at the Institut d’Astrophysique de Paris (CNRS). I took up my current position as an Eric and Wendy Schmidt AI in Science Postdoctoral Fellow in Oxford and a Nicholas Kurti Junior Research Fellow in 2025. I also hold the International Astronomical Union The Gruber Foundation (TGF) Fellowship in Astrophysics for 2025/26.

I teach the third year Physics course on symmetry and relativity (B2) at Brasenose.

My research focuses on developing statistical and machine learning methods for cosmology and astrophysics. As new surveys map the Universe in ever-greater detail, our models must evolve to capture both the complexity of large-scale structures and the physics of galaxy formation. Machine learning offers a way to bridge the gap between theory and data at scale, but its “black box” nature raises important questions of trust and interpretability. I work on building high-performance ML models that remain faithful to physical principles, enabling robust discoveries about dark matter, dark energy, and galaxy evolution.

A central part of my work is symbolic regression (SR), the task of automatically discovering analytic equations directly from data. Unlike deep neural networks, SR produces compact, interpretable expressions that reveal the underlying structure of a system, extrapolate reliably, and run orders of magnitude faster. I develop statistically principled approaches to SR, such as incorporating prior knowledge with language models, and design symbolic emulators that achieve the accuracy of neural networks while remaining lightweight, transparent, and portable across applications.

Much of this research is carried out within large international collaborations. For example, I co-lead the Accelerated Forward Modelling working group in the Simons Foundation’s Learning the Universe collaboration, where we use machine learning to accelerate and refine simulations of cosmological and astrophysical processes. These efforts enable simulation-based inference on the next generation of galaxy surveys, providing new opportunities to probe the nature of dark matter, dark energy, and the formation of structure in the Universe.