We are seeking
a full-time Postdoctoral Research Assistant to join the Oxford Witt Lab for Trust in AI (OWL) in the Department of Engineering Science (Central Oxford). The post is funded by Open Philanthropy and is fixed-term until 31 Dec 2027. A further extension to this would be based on funding availability. You will conduct hands-on empirical research in multi-agent security and agentic AI security, focused on adversarial testing (red-teaming) and mitigation of hard-to-detect failure modes in interactive AI systems (e.g., covert communication, collusion, strategic behaviour). You will be responsible for developing and running adversarial evaluations; designing and validating mitigation strategies (system-level controls, training-time or post-training defences, and deployment safeguards); publishing results in leading venues; contributing documented software to the group library; and supporting supervision of student research projects. The postholder will have substantial ownership over one or more core research thrusts within the lab’s multi-agent security programme. You should possess a completed PhD/DPhil (or thesis submitted by the start date) in Computer Science, Machine Learning, AI, Security, Robotics, Statistics, or a closely related quantitative discipline; a track record of research output appropriate to career stage; strong empirical research skills; an adversarial/security mindset grounded in threat modelling; strong programming and research engineering ability (Python with modern ML tooling such as PyTorch/JAX) and good software practices; familiarity with interactive/agentic settings (e.g., tool-using LLM agents, reinforcement learning, multi-agent systems); and excellent communication and collaboration skills with a commitment to an inclusive, respectful research environment. For more information about working at the Department, see Only online applications received before midday on 2 March 2026 can be considered.