Advanced degree (Master's or PhD) in Computer Science, Machine Learning, or Statistics with research experience.
Proven applied research track record with published work, open-source contributions, or impactful ML products.
Deep understanding of machine learning fundamentals including classical and modern approaches.
Experience in building and deploying AI/ML systems from scratch.
Hands-on expertise in model evaluation and validation, focusing on model failure.
Proficient in Python and experienced with ML frameworks like PyTorch and TensorFlow.
Familiarity with Large Language Models and their unique safety and evaluation challenges.
Knowledge in AI safety, robustness, or adversarial ML is advantageous.
Responsibilities
Conduct technical research on AI model behavior and practical risk assessment.
Develop AI systems for assessing the safety and reliability of other AI models.
Create production-grade platforms for automated AI risk evaluation and monitoring.
Assess various AI systems, identifying insurable risks and potential failure scenarios.
Establish evaluation methodologies to address AI-specific risks such as adversarial vulnerabilities.
Collaborate with underwriting teams to convert technical findings into usable risk insights.
Stay updated on AI research and quickly prototype new risk assessment techniques.
Benefits
Pioneering the evaluation of AI systems in a new and growing field.
Impact both technical challenges and practical real-world applications.
Engage in the unique challenge of building AI that evaluates other AI.
Shape how AI risk is quantified and managed across various industries.
Experience a fast-paced startup culture where ideas are rapidly implemented.
Opportunity to grow expertise at the intersection of AI research and insurance.
Freedom to explore innovative research directions that benefit both the field and the company.
Full Job Description
About Armilla AI
Armilla AI is a cutting-edge startup at the intersection of artificial intelligence and insurance. Based in Toronto, Ontario, Canada, we're building innovative solutions to manage, underwrite, and insure the rapidly evolving risks associated with AI systems. We're a dynamic team passionate about pioneering the future of AI risk management.
The Role
We're seeking an exceptional Applied Scientist who bridges the worlds of deep AI research and practical, production-grade applications. As our Applied Scientist, AI Risk, you'll be instrumental in both advancing our understanding of AI systems and translating that knowledge into robust risk assessment and evaluation frameworks. This isn't just about building models-it's about understanding their failure modes, vulnerabilities, and real-world reliability. You'll develop AI systems that evaluate other AI systems, creating the next generation of automated risk assessment tools. You'll be shaping how the insurance industry evaluates and prices AI risk.
Role Responsibilities
Conduct deep technical research into AI model behavior, failure modes, and edge cases, with a focus on practical risk assessment as well as deep research.
Design and develop AI systems-including specialized models, agents, and automated evaluation pipelines-that assess the safety, reliability, and risk profiles of other AI systems.
Build production-grade tooling and platforms for automated AI risk assessment, model testing, and continuous monitoring.
Evaluate diverse AI systems-from traditional ML models to Large Language Models and multimodal systems-identifying insurable risks, vulnerabilities, and potential failure scenarios.
Develop novel evaluation methodologies and metrics that capture AI-specific risks such as adversarial vulnerabilities, distribution shift, hallucinations, and behavioral misalignment.
Collaborate closely with our underwriting and actuarial teams to translate technical findings into actionable risk insights and pricing signals.
Stay at the forefront of AI research and safety literature, rapidly prototyping and validating new risk assessment techniques.
Write clean, maintainable, and well-tested code that scales from research prototypes to production systems.
Communicate complex technical concepts clearly to diverse stakeholders, from engineers to underwriters to executive leadership.
What We're Looking For
Advanced degree (Master's or PhD) in Computer Science, Machine Learning, Statistics, or related field, with demonstrated research experience in AI/ML.
Strong track record of applied research-you've published, contributed to open source, or shipped ML products that had real-world impact beyond academic settings.
Deep technical expertise in machine learning fundamentals, including both classical ML and modern deep learning approaches.
Experience building AI/ML systems from conception to deployment, not just running evaluations on existing models.
Hands-on experience with model evaluation, testing, and validation-you think critically about where models fail, not just where they succeed.
Solid software engineering skills with expertise in Python and experience with ML frameworks (PyTorch, TensorFlow, JAX) and scientific computing libraries (NumPy, Pandas, Scikit-learn).
Experience with LLMs and generative AI, including familiarity with their unique risks, evaluation challenges, and safety considerations.
Background in AI safety, robustness, interpretability, or adversarial ML is a significant asset.
Ability to work both independently on deep technical problems and collaboratively in a fast-paced startup environment.
Intellectual curiosity about the "other side"-not just building AI, but understanding its risks, limitations, and societal implications.
Strong problem-solving abilities and attention to detail, with a pragmatic approach to balancing research rigor with business needs.
What's In It For You
Pioneering a New Frontier: You'll be at the forefront of an emerging field, defining how AI systems are evaluated, understood, and insured at scale.
Dual-Sided Impact: Work on both the technical frontier of AI evaluation and the practical challenge of real-world risk assessment-a rare combination.
Meta-AI Challenge: Tackle the fascinating problem of building AI systems that understand and evaluate other AI systems.
Impactful Work: Your research and tooling will directly shape how AI risk is quantified and managed across industries.
Startup Agility: Enjoy the fast-paced, innovative, and collaborative culture of a growing startup where your ideas can quickly become reality.
Professional Growth: Unparalleled opportunities to develop expertise at the intersection of AI research, risk management, and insurance alongside deeply experienced AI and industry experts.
Technical Freedom: Latitude to pursue novel research directions and evaluation approaches that advance both the field and our business.
