Neuroscience researcher — mechanistic interpretability and computational neuroscience.

I study how neural systems, artificial and biological, represent information and arrive at decisions. My work sits between mechanistic interpretability — reverse-engineering the internal computations of trained networks — and computational neuroscience — modelling the dynamics of neural circuits. The throughline is a single aim: turning opaque systems into ones whose behaviour can be explained, predicted, and reasoned about.

• Mechanistic interpretability of neural systems — identifying the internal features and circuits through which networks map inputs to decisions, and relating them to theories of neural coding. Feature attribution, circuit-level decomposition, sparse features, activation and representation probing, ablation.
• Computational neuroscience — modelling neural dynamics and connecting these models to measurable prediction, including neuroimaging-derived structure and brain–computer-interface signals.

interpretability · explainable AI · neural circuits · representation analysis · neural dynamics · neuroimaging · brain–computer interfaces