Research Associate in Statistics and Machine Learning,
Department of Mathematics,
Imperial College London
180 Queen's Gate
South Kensington
London, SW7 2AZ
Research
I'm a mathematician and computer scientist at Imperial College London. I'm interested in studying the internals of AI systems, such as large language models, to understand how they represent information and reason about the world. The fundamental goal of my research is to ensure that the AI systems that we build and rely upon are safe and aligned with human values.
AI interpretability research has a lot of parallels with neuroscience research, except instead of studying brains, we study a neural networks. Unlike brains, we can perform perfectly reproducible experiments on neural networks and measure exactly how they process and respond to any stimulus. From these experiments we can learn things about, for example, how language models represent information about the world, and the algorithms they use to manipulate these representations and reason about the world.
Through this research, I hope that we might be able to reliably detect when AI systems are misaligned, before they are deployed in the real world, and we might be able to make interventions to modify their behaviour when they are (performing artificial neurosurgeries, if you will). This field of research is still very young, and we are still very far from achieving this goal. It is also arguably very important: uncontrolled AI is frequently cited as a plausible existential risk to humanity.
I am also broadly interested in the mathematical foundations of statistics and machine learning, in particular relating to exploratory data analysis such as clustering, visualisation and manifold learning, and to matrix-valued and time-varying data such as networks.
I currently hold the position of Research Associate at Imperial College London, under the supervision of Nick Heard. In September, I start a new position as a Chapman Research Fellow. I am currently organising the Network Stochastic Processes and Time Series Seminar. I previously completed my PhD at the University of Bristol, entitled "Spectral embedding of large graphs and dynamic networks", supervised by Patrick Rubin-Delanchy.
