Contributor to GPT-5 and OpenAI o3 models; published several papers such as 'Forgetting Transformer: Softmax Attention with a Forget Gate' (ICLR 2025), 'Maxwell’s Demon at Work: Efficient Pruning by Leveraging Saturation of Neurons' (TMLR); presented work at top conferences including NeurIPS, ICLR.
Research Experience
Interned with David Silver's RL team at DeepMind during his PhD.
Education
PhD in Computer Science from Mila, University of Montreal, advised by Pierre-Luc Bacon and Aaron Courville; previously studied at Cornell University, Higher School of Economics, and Lomonosov Moscow State University.
Background
A Member of Technical Staff at OpenAI, building the next generation of reasoning models. His research interests include parameter, compute, and data efficiency in reinforcement learning.
Miscellany
Contact information includes email, CV link, and profiles on Google Scholar, Twitter, GitHub.