People

Iliad Fellow

Maths graduate of the Australian National University, from Canberra. Works on agent foundations and abstractions, and on black-box methods for studying AI systems. April 2026 Iliad Intensive alum.

MATS extension fellow

Works on agent foundations: self-models for embedded agents, and how coherent identities form in AIs and humans, drawing on behavioural game theory and computer science.

PhD student, Australian National University; Head of Teaching, ARENA

Works on universal AI with Marcus Hutter, and co-wrote the textbook An Introduction to Universal Artificial Intelligence.

Assistant Professor in Computer Science and AI, University of Sussex

Works on emergence, information theory and AI alignment, including high-order interdependencies in complex networks. Co-author of AI in a vat, on the limits of efficient world modelling for agent sandboxing and interpretability.

Research scientist, Stormglass (formerly Moirai)

Works on deep learning theory, training dynamics and interpretability, including SGD dynamics and phase transitions in deep learning. PhD from the University of Sussex on information flow between brain regions during visual perception.

Visiting researcher, Meridian Cambridge

Works on encoded reasoning, steganography and chain-of-thought monitoring. Co-author of Steering Llama 2 via Contrastive Activation Addition. Wrote and taught Iliad's ML engineering and mechanistic interpretability days.

University of Oxford

Co-wrote the Iliad Intensive's singular learning theory materials, with Matthew Farrugia-Roberts and Zach Furman.

Fall 2026 Iliad fellow

Recent PhD in Extremal and Probabilistic Combinatorics from UCL. Worked as a data scientist and studied at Oxford and Edinburgh before.

Member of Technical Staff, EleutherAI

Works on interpretability, deep learning theory and data attribution. Co-author of Bergson, an open-source library for data attribution. Mathematics PhD from Imperial College London, in arithmetic geometry.

Iliad Fellow

Works on the mathematical foundations of intelligence, including natural latents and whether natural abstractions survive shifts in ontology.

Iliad staff

Master's in physics, on quantum information, and a PhD in mathematics, on provable interpretability and multi-agent communication. Ask him about subliminal learning, prover-estimator debate, steganography and undetectable backdoors.

Professor of Geometry and Physics, University of Edinburgh

Works on the mathematics of quantum field theory and string theory, and on bridging the two. Iliad Fellow, interested in a scientific theory of deep learning.

Simplex

Works on computational mechanics and belief-state geometry in language models. Co-author of Large Language Models Develop Belief State Geometry In-Context.

Research Engineer, Timaeus

Works on interpretability and machine learning engineering. Author of Compressed Computation: Dense Circuits in a Toy Model of the Universal-AND Problem, and co-author of Spectroscopy at Scale, a study of interpretable structure in Pythia-1.4B.

University of Bonn

Mathematical physicist working on quantum field theory, general relativity and quantum gravity, in particular Hopf-algebraic renormalisation and BV-BRST quantisation. Now applying those methods to AI search and reasoning.

University of Washington; Fall 2026 Iliad fellow

Mathematician who classifies symmetries, or more precisely the ambient spaces that can host them, and argues that algebraic geometry can and should be built beyond (super) vector spaces.

Universitat Politècnica de Catalunya

Recent graduate in mathematics and physics from UPC in Barcelona, with a final year at Oxford, and two IMO bronze medals. Interested in singular learning theory and agent foundations.

Iliad Fellow

MSc in theoretical physics from the Netherlands and MSt in philosophy of physics from Oxford. AI safety projects with SPAR and Dovetail; April 2026 Iliad Intensive alum. Also worked on building a satellite that is now in orbit.

DPhil student, University of Oxford

Works on singular learning theory, goal misgeneralisation and the foundations of reward learning. Co-author of Loss Landscape Degeneracy and Stagewise Development in Transformers.

University of Oslo; Fall 2026 Iliad fellow

PhD in theoretical chemistry, on numerical methods for the Schrödinger equation. Interested in developmental interpretability, and in the right level of abstraction for describing a neural network.

Iliad Fellow

Mathematics graduate of the Australian National University, with an honours thesis in representation theory. Works where deep learning theory meets interpretability.

PhD student, University of Melbourne

Works on singular learning theory and the mathematical foundations of deep learning. Co-author of The Local Learning Coefficient: A Singularity-Aware Complexity Measure, and of Benign Loss Landscapes Can Coexist with Worst-Case Hardness.