Scholar
Thomas Jiralerspong
Google Scholar ID: ifQsJnIAAAAJ
Graduate Student - Computer Science (AI option) - Mila/Université de Montréal
Citations & Impact
Citations
170
H-index
6
i10-index
4
Publications
20
Co-authors
0
Publications
Resume (English only)
Academic Achievements
- Chosen as one of 200 most promising young researchers in math & CS by the Heidelberg Laureate Forum; Won the Project X ML research competition (AI for healthcare); Submitted two papers to ICLR 2026: 'Cross-Architecture Model Diffing With Crosscoders' and 'Learning What Matters: Steering Diffusion via Spectrally Anisotropic Forward Noise'.
Research Experience
- Co-founder, CTO, and CSO of Lucent Cell; Research fellow at Anthropic (mechanistic interpretability); Research scientist at Occam AI (LLM agents); Research intern at Waabi (autonomous driving); Research intern at the Vector Institute for AI (AI for HVAC); Research intern with Blake Richards (neuroAI); Research intern with Doina Precup (model-based RL); Software development intern at Amazon, Expedia, and Square Enix.
Education
- PhD student in artificial intelligence at Mila and Université de Montréal, co-supervised by Yoshua Bengio and Guillaume Lajoie. Research supported by Vanier, NSERC, and FRQNT scholarships.
Background
- Research interests include mechanistic interpretability, compositionality, causality, LLM agents, autonomous driving, healthcare ML, neuroAI, AI for HVAC, reinforcement learning, and diffusion models. The primary research goal is to understand and build safe AI systems and use them to solve real-world problems.
Miscellany
- Enjoys making music, traveling the world, and playing strategy/deception games.
Co-authors