Bio

Last updated 2026-09-12

I studied a Bachelors of Science (Advanced Mathematics), double-majoring in Mathematics and Computer Science at the University of Sydney, including an Honours year in Applied Mathematics (translation for non-Australians: this is an additional research year tacked onto the end of an undergraduate degree). My Honours thesis was on Discrete Morse Theory.1

I then worked for ~5 years, first at Atlassian (in their Ecosystem team), and then at Canva (in their Trust & Safety team). I had a great time and learned a lot about how to solve real-world problems with data at scale, and how to get stuff done.

However, the academic itch never left, so I studied a MSc in Social Data Science at the University of Oxford. I was lucky enough to get into the DPhil in Social Data Science, supported by the Shirley Scholarship (from the OII) and the BOLD Lab. I will start in Oct 2026, and am part of Exeter College.

Effective altruism has influenced me for a long time, primarily because it takes seriously the imperative for people to do good with their life (especially those with more resources), and to try to do so in a carefully considered manner. I think AI is likely to be transformative. It is important for it to go well, and I am concerned that capability gains are outstripping safety gains.

Some quick takes:

  • My theory of change is that independent2 and high-quality safety research and evaluations can a) make a co-ordinated slowdown more likely, b) make AI safer, and c) improve the quality and efficacy of regulations.
  • I do not think that addressing already-realised harms and possible future harms are mutually exclusive or in conflict except in some rare edge cases.
  • I think that the ‘we are in a great power struggle with China and thus need to kick off the arms race and throw safety to the wind’ argument is dangerous for many reasons, and I have been quite shocked by the seeming consensus among American elites (of all political persuasions) on this point.

Contact

  • Personal: {firstname}.{lastname}98@gmail.com
  • Oxford email: {firstname}.{lastname}@oii.ox.ac.uk

  1. This provides a way to meaningfully define a gradient flow on a graph (yes, nodes and edges) so that powerful methods from smooth manifolds transfer across directly! If you have been successfully nerdsniped and have a maths background, check this out. ↩︎

  2. I am wary of the maximalist and overly big-brained logics that can lead people to work inside frontier labs. ↩︎