Hello there — I’m an undergraduate researcher at Rinn AI (foremly, Insight SFI Centre) and a BSc student in Data Science & AI at University College Cork.

I’m interested in world models, in AI for physics and mathematics, and in quantum computing. The question underneath all three, for me, is how much a system can learn about the world it operates in by acting and measuring, rather than by being told.

Currently, I lead the Tool World Model project, a collaboration between UCC and IBM Research. An agent that uses software tools is told what each tool does, but not how the tools fit together, and documentation leaves more of that unsaid than you would expect. We’re building an agent that learns the missing structure by using the tools. We want two things from it: to work in a suite nobody has documented, and to find its own ways of getting things done. If you draw a tool suite as a graph, with tools as nodes, a workflow is just a subgraph, and there is no reason the useful ones are only those a person would think of. Finding the ones nobody has written down, or would have, is the part I find most interesting. Alongside this I built HERMES, a differentiable pipeline for finding planets in Kepler light curves, and used it to test what a physics-informed feature really contributes.

Longer term, I want to bring these threads together: world models grounded in physics, for planning in embodied systems and for scientific discovery.

My CV can be found here.