ML Research Scientist - Leading AI Lab, Montreal

About

We're partnering with a highly respected, mission-driven AI research lab tackling one of the field's most important long-term challenges: building AI systems that are fundamentally safe by design.

The Role

We're seeking an ML Research Scientist to contribute to foundational work on safe AI systems. This is a generalist research position focused on developing principled, technically rigorous approaches to AI safety — centered on probabilistic guarantees, structured reasoning, and uncertainty modelling rather than post-training mitigation techniques.

What You'll Do

• Conduct foundational research on probabilistic safety guarantees for large-scale AI systems
• Develop new methodologies for reasoning, causality, robustness, and generalisation
• Advance work on uncertainty quantification and reliability in language models
• Contribute to distributed training and evaluation of large-scale models under safety constraints
• Design rigorous experimental frameworks to test and validate safety-oriented approaches
• Publish high-impact research advancing the field of AI safety
• Work closely with a small team of world-class researchers operating at the intersection of theory and large-scale experimentation

What We're Looking For

• PhD in Computer Science, Physics, Mathematics, or related field (or equivalent research depth in industry)
• Strong publication record in leading venues (e.g., NeurIPS, ICML, ICLR, COLT, ACL)
• Deep experience with language models and large-scale training pipelines
• Strong grounding in areas such as probabilistic modelling, Bayesian inference, statistical learning theory, causality, robustness, or uncertainty estimation
• Solid mathematical foundations combined with production-level coding skills (Python; PyTorch/JAX experience preferred)
• Experience designing and running large-scale experiments with rigorous evaluation
• Pragmatic approach to AI safety research (not ideological)
• Collaborative mindset — low ego, team-first mentality

Why This Matters

This lab operates independently without commercial pressure cycles, allowing full focus on rigorous, long-horizon research. You'll work alongside some of the best researchers globally who are contributing to technically serious solutions that could shape the future of AI safety.

Location

Montreal (onsite) - open to candidates willing to relocate