Alexander D. Goldie

Alexander D. Goldie

I am a PhD student at the University of Oxford, under the supervision of Jakob Foerster and Shimon Whiteson. My research focuses on automated algorithm discovery, autonomous AI research and meta-reinforcement learning.

I will soon be starting a research internship at Meta! Before that, I was a Research Scientist Intern at Wayve, where I worked on offline reinforcement learning for autonomous driving. My PhD is offered by AIMS, a competitive PhD-level course for Machine Learning.

Email / CV / Scholar / Twitter / Github / LinkedIn

News

(06/26) I start a Research Scientist Internship at Meta!
(06/26) I was invited to speak about DiscoGen at the first AIDDA Conference.
(05/26) DiscoGen was accepted to ICML 2026!
(05/26) I was awarded Gold Reviewer at ICML 2026.
(04/26) I won a grant worth $240k from OpenPhilanthropy, on behalf of the lab, for scaling faithfulness in chain-of-thought reasoning.
(03/26) I gave an invited talk about DiscoGen at Cambridge University (CaMLSys Lab).
(02/26) I gave the inaugural Cruickshank Lecture about DiscoGen at Inherent Labs.
(12/25) We released a blog for the initial version of DiscoBench (a precursor to DiscoGen).
(11/25) My internship at Wayve finished. During my internship, for a while, I developed the model which was driving the Wayve car!
(08/25) I was invited to talk on two episodes of TalkRL.
(08/25) My paper was awarded "Outstanding Paper for Scientific Understanding In Reinforcement Learning" at RLC 2025!
(06/25) I started a Research Scientist internship at Wayve.
(12/24) OPEN was awarded a Spotlight at NeurIPS 2024!
(07/24) I took part in a panel discussion at the AutoRL workshop in ICML24.
(06/24) OPEN was awarded a Spotlight at the AutoRL workshop in ICML24.
(10/22) I started my PhD on AIMS at the University of Oxford.
(07/22) I graduated from my MEng Engineering Science at the University of Oxford with a high First!

Research

	DiscoGen: Procedural Generation of Algorithm Discovery Tasks in Machine Learning Alexander D. Goldie, Zilin Wang, Adrian Hayler, Deepak Nathani, Edan Toledo, Ken Thampiratwong, Aleksandra Kalisz, Michael Beukman, Alistair Letcher, Shashank Reddy, Clarisse Wibault, Theo Wolf, Charles O'Neill, Uljad Berdica, Nicholas Roberts, Saeed Rahmani, Hannah Erlebach, Roberta Raileanu, Shimon Whiteson, Jakob N. Foerster ICML 2026 paper \| site \| code \| tweet
	Evolution Strategies At The Hyperscale Bidipta Sarkar, Mattie Fellows, Juan Agustin Duque, Alistair Letcher, Antonio León Villares, Anya Sims, Clarisse Wibault, Dmitry Samsonov, Dylan Cope, Jarek Liesen, Kang Li, Lukas Seier, Theo Wolf, Uljad Berdica, Valentin Mohl, Alexander D. Goldie, Aaron Courville, Karin Sevegnani, Shimon Whiteson, Jakob N. Foerster ICML 2026* paper \| code \| site \| tweet
	An Optimisation Framework For Unsupervised Environment Design Clarisse Wibault, Alexander D. Goldie, Antonio Villares, Maike Osborne, Jakob N. Foerster. DEMO Workshop @ RLC 2026 arXiv \| site \| tweet
	Model-Based Meta-Learning for Algorithm Discovery Theo Wolf, Alexander D. Goldie, Jarek Liesen, Uljad Berdica, Mattie Fellows, Jakob N. Foerster World Models Workshop @ ICLR 2026 paper
	Learning To Drive in New Cities Without Human Demonstrations Zilin Wang, Saeed Rahmani, Daphne Cornelisse, Bidipta Sarkar, Alexander D. Goldie, Jakob N. Foerster, Shimon Whiteson SAD Workshop @ CVPR 2026 arXiv \| code \| site \| tweet
	How Should We Meta-Learn Reinforcement Learning Algorithms? Alexander D. Goldie, Zilin Wang, Jaron Cohen, Jakob N. Foerster, Shimon Whiteson RL Conference 2025 (Outstanding Paper) arXiv \| code \| tweet
	An Optimisation Framework For Unsupervised Environment Design Nathan Monette, Alistair Letcher, Michael Beukman, Matthew T. Jackson, Alexander Rutherford, Alexander D. Goldie, Jakob N. Foerster RL Conference 2025 arXiv \| code \| site \| tweet
	Can Learned Optimization Make Reinforcement Learning Less Difficult? Alexander D. Goldie, Chris Lu, Matthew T. Jackson, Shimon Whiteson, Jakob N. Foerster NeurIPS 2024 (Spotlight Award) Also at AutoRL Workshop @ ICML2024 (Spotlight) arXiv \| code \| tweet
	Adam On Local Time: Addressing Nonstationarity In RL With Relative Adam Timesteps Benjamin Ellis, Matthew T. Jackson, Andrei Lupu, Alexander D. Goldie, Mattie Fellow, Shimon Whiteson, Jakob N. Foerster NeurIPS 2024 paper
	Robust Offline Learning via Adversarial World Models Uljad Berdica, Kelvin Li, Michael Beukman, Alexander D. Goldie, Mattie Fellows, Perla Maiolino, Jakob N. Foerster Open-World Agents Workshop & Adversarial ML Workshop @ NeurIPS24 paper