Postdoctoral Research Associate

The role is part of AI²: Assurance and Insurance for Artificial Intelligence, a major research programme on insuring AI systems. The project is a collaboration between the University of Edinburgh, the University of Oxford, the University of Warwick and AXA. It aims to develop the mathematical, statistical and computational foundations needed to evaluate, price, monitor and mitigate risks arising from AI systems, including generative and agentic AI.

The successful candidates will work on rigorous methods for AI risk assessment and insurance. Relevant research areas include risk measures, systemic risk, conformal prediction, uncertainty quantification, stochastic and Bayesian optimisation, reinforcement learning, AI security, red-teaming and jailbreak evaluation. Candidates are not expected to cover all of these areas, but should have strong expertise in at least one and an interest in developing methods that apply to AI assurance, risk management and insurance.

The Opportunity:

The School of Mathematics at the University of Edinburgh is at the forefront of research in financial mathematics, stochastic analysis, machine learning, AI safety and quantitative risk. The successful candidates will join a growing interdisciplinary team working on the foundations of AI insurance and algorithmic assurance.

The project will develop tools and theory for questions such as: how should AI-related risk be measured; how can uncertainty and tail risk be quantified; how can failures of AI systems be modelled and priced; how can systemic risk emerge from interacting AI systems; and how can testing, validation, monitoring and red-teaming evidence be translated into insurance and risk-management frameworks?

The posts offer the opportunity to work closely with leading academic and industry partners, including AXA, and to contribute to a broader international research agenda on safe, trustworthy and insurable AI.

Main responsibilities

The postholders will be expected to:

• Conduct high-quality research on AI risk, algorithmic assurance and insurance
• Develop mathematical, statistical or computational methods for evaluating, pricing and monitoring AI-related risks
• Work on one or more relevant areas, including risk measures, systemic risk, conformal prediction, stochastic/Bayesian optimisation, reinforcement learning, AI security, red-teaming, jailbreak evaluation or agentic AI evaluation
• Build reproducible research code, simulations, benchmarks or evaluation tools where appropriate
• Publish research in leading peer-reviewed journals and conferences
• Present findings at academic conferences, industry workshops and project meetings.
• Collaborate with researchers at Edinburgh, Oxford and Warwick, and with industry partners including AXA.
• Contribute to the development of the broader AI² research agenda.
• Support project activities, including workshops, reports, grant development and engagement with stakeholders in insurance, finance, AI assurance and regulation.