Postdoctoral Research Associate in Probabilistic AI Safety and Rare-Event Simulation in London

Are you a quantitative researcher interested in the foundations of AI safety and reliability? We are seeking a Postdoctoral Research Associate to join PRISM (Probabilistic Rare-event Inference for Safety of Models), a new research programme developing rigorous statistical methods to quantify extremely rare but high-impact failures in large language models and other generative AI systems. Working in the Department of Mathematics at Imperial College London, you will be at the forefront of transforming AI safety evaluation into a reproducible, auditable science grounded in probability, statistics, and large-scale computation.

In this role, you will play a central part in the design, implementation and validation of new probabilistic methodologies for AI safety assurance. Your work will combine theory, algorithms and computation, with clear pathways to real-world impact. Specifically, you will:

• Develop and analyse rare-event simulation and Sequential Monte Carlo methods tailored to generative AI systems.
• Design and run large-scale computational experiments to estimate extremely low-probability unsafe behaviours in language models.
• Contribute to the development of an end-to-end research pipeline, integrating prompt generation, stochastic rollouts, verification, and uncertainty-quantified risk estimates.
• Produce high-quality research outputs, including peer-reviewed publications, open-source software, and technical reports.
• Collaborate closely with academic colleagues, research software engineers, and external partners working at the interface of statistics, machine learning and AI governance.
• Present your findings at seminars, workshops and international conferences, and contribute to the wider intellectual life of the research group.

You will be a motivated and intellectually curious researcher with strong quantitative foundations and an interest in challenging, open-ended problems. We are particularly interested in candidates who have:

• A PhD (or near completion) in Statistics, Machine Learning, Applied Mathematics, Computer Science, or a closely related discipline.
• A strong background in probability, statistics, or statistical inference.
• Experience conducting independent research and producing high-quality outputs (publications, preprints, software).
• Strong programming and computational skills, for example in Python and scientific computing or machine-learning frameworks.
• The ability to communicate complex technical ideas clearly, both in writing and verbally.
• A collaborative mindset and the ability to organise your own work effectively.
• Experience with Monte Carlo methods, generative models, or AI evaluation is desirable but not essential.

The opportunity to work on a high-profile, methodologically ambitious research programme at the forefront of AI safety and statistical science. The chance to develop foundational research with relevance to industry, regulation and public policy. The opportunity to continue your career at a world-leading institution and be part of our mission to continue science for humanity. Grow your career: gain access to Imperial’s sector-leading as well as opportunities for promotion and progression. Sector-leading salary and remuneration package (including 41 days off a year and generous pension schemes). Be part of a diverse, inclusive and collaborative work culture with various resources to support your personal and professional development.