Welcome!

1) Overview

I am a Canadian AI graduate student finishing my second M.Sc. in Computer Science at the Mila – Quebec AI Institute. My master’s research focused on representation learning for deep reinforcement learning with Pablo Samuel Castro as my advisor, with whom I co-authored a survey published in TMLR. Separately from my master’s work, I additionally collaborate with Aaron Courville, Hugo Larochelle, Juan Duque, and other Mila researchers on multi-agent reinforcement learning in social-dilemma settings, for example through opponent-shaping methods. Our first work in this direction studies cooperation and stability in social-dilemma settings under climate and economic constraints.

On the industry side, I had the chance this year to be involved in deep learning and reinforcement learning projects for engineering R&D at Bombardier Aerospace and Bentley Systems. At Bombardier, I developed AI/data pipelines for flight-test data to train DNNs for indirect airspeed estimation and performance modeling. At Bentley, I worked on exploring ways to apply deep reinforcement learning to their specific problem of automated civil asset labeling on engineering drawings. Overall, I’m grateful for these opportunities, which showed me both the challenges and potential of large-scale AI and automation adoption across engineering tasks in the design cycle.

2) Research Interests

- Sensing, Reasoning, and Control:
• Representation learning from multimodal and high-dimensional data
• Deep Reinforcement Learning for complex policy learning and control
• Predictive world modeling, multi-step reasoning, and hierarchical planning

- Theoretical Challenges:
• Fundamental aspects of learning and intelligence
• Achieving better sample efficiency and generalization
• Optimizing target performance under resource constraints

- Applied Problems:
• AI-enhanced scientific simulations (e.g., CFD/FEA with surrogate models)
• Robotics learning in complex simulators, sim2real and real2sim challenges
• End-to-end learning for combinatorial problems (e.g., ressource allocation)
• AI-based decoding and adaptation methods for future Brain–Computer Interfaces
• Autonomous aircraft & spacecraft control using hybrid RL/MPC-based approaches
• Multi-Agent RL for Policymaking Research and Applications (e.g., space debris issue)

3) Personal Interests

I enjoy visiting technology, science, and space museums while traveling, having driven over 1.5 times the Earth’s circumference across my last three U.S. road trips for that purpose. I have a broad interest in reading across many subjects and enjoy practicing various physical sports.