Hi there! I’m Ziyan “Ray” Luo, a Ph.D. candidate at Mila, McGill, fortunate to work with (and learn from) Dr. Xujie Si, Dr. Doina Precup, and my talented colleagues, who continuously inspire me.

🌟 Research Interest: RL & Abstraction

I am broadly interested in Reinforcement Learning (RL), currently with a focus on representation learning—developing compact, structured, and agent-centric encodings that support generalization, sample-efficient learning, and planning. As RL scales to complex domains like robotics, such representations become essential for tractable and robust decision-making.

A central theme in my research is understanding how abstraction can serve as a foundation for representation learning. Inspired by early work in formal verification, which emphasized abstraction as a tool for managing complexity through rigorous equivalence and refinement, I explore principles such as behavioral metrics for state abstraction and temporal abstractions in hierarchical RL as tools to integrate formal structure with learning-based flexibility.

I am also inspired by the ethos of the formal methods community: their emphasis on precision, scientific rigor, and long-term research impact shapes the way I approach problems.

🎼🏸 Beyond the Lab: Music, Sports & More

When I’m not exploring algorithms, you may find me composing music that tells stories. Here’s my portfolio (some tracks even made it into popular video games!). I love blending electronic sounds with acoustic instruments to create immersive, theme-driven pieces. Glad that they are well-received: now, I have over 10,000 followers in multiple Chinese music distribution platforms / social media!

Ball sports keep me energized: badminton, tennis, table tennis, billiards—you name it! And if you’re an animal lover, check out my Instagram for some furry friends.