Olly grew up in Wiltshire and studied mathematics and computer science at Christ Church, University of Oxford. During his final year, he worked on evaluating defences against distillation attacks and studied how a non-stationary training distribution affects memorisation and generalisation in small transformers. His dissertation developed theory concerning approximations to the singular values of large matrices with applications to neural network optimisers like Muon. During an internship at Quansight Labs, he added adaptive quadrature for multidimensional integration to SciPy, the Python scientific computing library. At StatML, Olly will work in collaboration with Meta on statistical methodology and principled evaluation procedures for modern AI systems, with a focus on cybersecurity. In his free time, he enjoys long-distance running, travel and live music.
Olly Britton
