About me

I did an undergraduate degree in computer science at Imperial College London and my PhD in machine learning at Oxford University (2021), supervised by Profs. Andrea Vedaldi and Joao Henriques in the Visual Geometry Group. During the PhD I developed a general purpose information theoretic clustering algorithm for neural networks that became widely adopted. After the PhD I did a postdoc at Mila supervised by Prof. Yoshua Bengio (2023) working on fundamental principles of neural network learning, using information theory to shine a light on how neural networks generalise to new inference datapoints. Highlights of my work at Mila include the first proof of a generalization bound for the information bottleneck principle since its inception by Naftali Tishby in 2001, establishing equivalences between variational generative models, and an interdisciplinary paper using Kolmogorov information theory to bound ineffability in communication. In 2024 I deepened my spiritual practice, becoming certified in teaching meditation and yoga and releasing a book on stoic affirmations, and in mid 2024 I joined the Calico-funded Westminster-Imperial-Oxford consortium implementing and analysing imaging biomarkers for the UK Biobank, motivated by a desire to combine health and informatics to improve disease understanding and prevention. Currently I am working on formulation of biological age metrics and discovery of mechanisms underlying aging and age-associated disease.