I develop computational and conceptual foundations for understanding the changing world. Complex systems, whether biological, ecological, or societal, are in constant flux, and a central challenge of our time is learning to reason about them rigorously: identifying patterns of change, distinguishing signal from noise, and building computational methods that work when stability cannot be assumed.
A central line of my research asks how AI systems can support rigorous reasoning in science and other high-stakes contexts. When a machine learning model mediates inference to produce a scientific quantity, support a consequential decision, or shape how evidence is interpreted, what formal conditions must it satisfy? I study learned measurement, robustness, fairness, and epistemic responsibility in AI systems, aiming to build the formal foundations that trustworthy AI use in reasoning requires. This work spans scientific inference, decision support, and the general conditions under which computational tools amplify rather than distort human reasoning.
I investigate macroevolutionary and macroecological dynamics using fossil and other long-term data, including diversification, expansion, persistence, and extinction at geological timescales, and how ecological relationships shape these processes. This work has produced findings on large-scale evolutionary patterns published in Nature, PNAS, and Nature Communications. A broader extension of this research, pursued through the HAT consortium which I lead, asks whether analogous patterns of change operate across other complex systems: ecological, cultural, linguistic, and societal. Do species, economies, languages, and cultures age in the same way? Can evolutionary intuitions developed over long timescales transfer to understanding rapid change in human systems? These questions connect evolutionary biology to complexity science and motivate the computational methods I develop.
I develop methods for studying systems where change is the phenomenon of interest. The research includes modelling evolving data streams, detecting and characterizing change processes, and evaluating computational methods when stability cannot be assumed. This work originates in computer science and is driven by problems where the questions set the agenda.
Across all three lines runs a single question: how do we reason carefully about a world that does not stand still? The evolutionary biology work is where I encountered the foundational problems; the methods work provides the tools; and the AI for reasoning work generalizes what rigorous computational reasoning requires across domains, instruments, and scales of change.