Job description
Mila • Montreal, Canada
Montreal, Canada
- In this role, you'll design and build an open-source security evaluation platform for AI agents.
- Implement and evolve a scoring engine with deterministic, embedding-based, and LLM-as-judge checks.
- Architect reproducible and auditable evaluation schemas and design policy enforcement infrastructure.
- Reason about threat models and attack surfaces to guide architecture decisions and collaborate with researchers.
- Mentor junior developers and maintain awareness of emerging AI security and adversarial techniques.
Mila is a research institute in Montreal, Quebec which mainly focuses on machine learning research.