About Me

I am an AI Research Scientist at Meta Superintelligence Labs where I work on post-training and evaluating LLMs for continual improvement based on large-scale online user interactions. My current research focus is on self-improving LLMs and multi-agent reinforcement learning for LLMs. Before joining Meta Superintelligence Labs, I was an AI Research Scientist at Meta's Ranking and Foundational AI Research group, working on large-scale generative learning for recommender systems. I was also a Machine Learning Lecturer at the University of Toronto teaching Fundamentals of Deep Learning. Prior to that, I held the position of Principal AI Research Scientist and Research Manager at the Autodesk AI Lab. I earned my PhD from the University of Ottawa, with a focus on deep learning for common-sense reasoning in virtual environments. My research has been published in top-tier AI venues including NeurIPS, ICLR, ICML, and TMLR.

Research Interests

Self-Improving LLMs · LLM Reasoning · Multi-Agent RL · Post-Training · Generative Retrieval

News

• [Jun 2026] One paper accepted at ICML 2026: Structure Enables Effective Self-Localization of Errors in LLMs
• [Mar 2026] Two papers accepted at EACL 2026: Imbalanced Gradients in RL Post-Training and How to Make LLMs Strong Node Classifiers?
• [2025] One paper accepted at KDD 2025: Generating Long Semantic IDs in Parallel for Recommendation
• [2025] One paper accepted at ICLR 2025: Learning Graph Quantized Tokenizers
• [2025] Joined Meta Superintelligence Labs