Bio

I am a PhD student at the University of Cambridge, supervised by Prof. Jamie Vicary and Prof. Pietro Liò. My research focuses on the role of structure in deep learning. I study the kind of structures (geometric, algebraic, logical) that emerge in neural representations, and I develop novel methodologies to prime networks to align with the structure of the data. Recently, I have also been studying transformer-based LLMs from an information-theoretic perspective, with applications to memorisation and reasoning/CoT.

Before my PhD, I completed an MPhil in Advanced Computer Science at the University of Cambridge, where I took courses in machine learning and category theory. My dissertation, Towards Category-Theoretic Message Passing, investigated category-theoretic and algebraic methods for enabling reasoning and compositionality in graph neural networks.

Prior to that, I obtained a BSc in Mathematics and Computer Science from the University of Manchester, where I studied pure mathematics, machine learning, and theoretical computer science. While at Manchester, I worked with Prof. Gavin Brown on learning theory, using information geometry to formalise the connection (and distinction) between the seminal notions of bias/variance and approximation/estimation.