I completed my PhD at Mila in March 2024, supervised by Aaron Courville and Marc Bellemare, working at the intersection of scaling and sample-efficient reinforcement learning. I am currently a Member of Technical Staff at Anthropic. Previously, I was VP of Research at OpenAI, where I led the post-training team, was a Member of Technical Staff on the Strawberry team (responsible for o1-preview), and led post-training for o1 and o3.

News

• March 2026: I’ve joined Anthropic as a Member of Technical Staff.
• September 2025: I’ve been promoted to VP of Research at OpenAI.
• March 2025: I’ve taken on the role of leading the post-training team at OpenAI.
• November 2023: I’m excited to be joining OpenAI, where I’ll be working on pushing the frontier of AI capabilities.
• September 2023: My internship at Apple in Samy Bengio’s machine learning research group has come to an end. It was a wonderful experience — thanks to the team!
• June 2023 I’m moving back to California to start an internship in Samy Bengio’s machine learning research group at Apple. See you all in the Bay!
• June 2023: I’m thrilled to announce that our new paper “Bigger, Better, Faster: Human-level Atari with human-level efficiency,” to ICML 2023. Reaching human-level sample-efficiency with pure model-free reinforcement learning on Atari 100k allows me to finally end my love-hate relationship with this benchmark and move on to a new phase in life.
• June 2023: Sadly, my time as a Student Researcher at Google Brain (now Google DeepMind) has come to an end. It’s been an incredible journey, and I’m grateful for the wonderful community at Google Montreal, inside and outside of research.
• February 2023: We’ve had two papers accepted at ICLR 2023! Our work Sample Efficient Reinforcement Learning by Breaking the Replay Ratio Barrier, on a new approach to scaling model-free RL, received a top 5% award, and our paper Simplicial Embeddings in Self-Supervised Learning and Downstream Classification received a top 25% award. Thanks to all of my coauthors!
• September 2022: Our paper, “Beyond Tabula Rasa: Reincarnating Reinforcement Learning,” has been accepted at NeurIPS 2022. This work highlights the importance of efficient retraining in reinforcement learning, mitigating the massive computational requirements of the field and allowing for
• June 2022: I’m honored to have received the Borealis AI Fellowship this year. It’s awarded to ten promising Canadian AI PhD students, and I’m humbled to be among them.
• May 2022: I’m excited to share that I’ve received the FRQNT PhD Scholarship, a full fellowship awarded to promising PhD students studying science and technology in Québec.
• May 2022: I’m happy to announce that our paper, “The Primacy Bias in Deep Reinforcement Learning,” will be published at ICML 2022. This work presents a novel and highly unusual study into the pathologies of deep reinforcement learning, and suggests a set of unintuitive remedies (iterative resetting agents’ parameters) may dramatically improve their performance and robustness.
• December 2021: I’m thrilled to share that our paper, “Deep Reinforcement Learning at the Statistical Precipice,” received the Outstanding Paper Award at NeurIPS 2021. It’s an honor to have our work recognized in this way, and I’m grateful to the community for this recognition.
• September 2021: Our papers Pretraining Representations for Data-Efficient Reinforcement Learning and Deep Reinforcement Learning at the Statistical Precipice have been accepted to NeurIPS 2021!