Research

For my PhD in computational neuroscience (University of Nottingham, 2020–2024), supervised by Mark Humphries and Matias Ison, I worked on inferring the wiring between neurons from their voltage recordings. Neuroscientists can now record precise voltages of many individual neurons in living animals; if we can infer the synaptic connections from such data, we could construct connectomes in vivo. In simulations, I developed three new connection tests; two of them (template correlation, and linear regression of the voltage upstroke) detect connections significantly better than the standard spike-triggered-average test. The thesis is called From Voltage to Wiring: Synaptic connectivity inference from neural voltage recordings (PDF). This YouTube video is a short presentation about the work.

Almost every figure in the thesis links to the Jupyter notebook that made it: all ~100 research notebooks are online, as a log of the whole project and a place to discuss results with my supervisors. Dan Goodman, one of my examiners, liked the setup enough to ask me to build the same for his COMOB project: an open, massively collaborative project modelling sound localization with spiking neural networks. I set up its website and notebook infrastructure. Resulting paper (eNeuro, 2025): Spiking Neural Network Models of Interaural Time Difference Extraction via a Massively Collaborative Process.

Along the way I built and optimised a spiking neural network simulation library in Julia, was a TA for the COSYNE 2022 tutorial on spiking neural networks (and ran technical infrastructure for it and for COMOB), and gave a Julia for Scientists talk.

After that I did a postdoc in neuroinformatics at Ghent University (2024), estimating Bayesian inverse models on large brain-imaging datasets, within the European EBRAINS project.