Open positions at TSG Lab are listed below. For instructions on how to get in touch — including what to include in your email and subject line codes — see our Contact page.
AI systems are increasingly deployed in ways that allow their behaviour to change after release—through in-context learning, memory mechanisms, or post-deployment fine-tuning. Most interpretability work treats models as frozen artefacts and analyses a snapshot. We're interested in what happens to the internals as a model learns from experience: how representations shift, which circuits are stable, and whether you can forecast where a system's behaviour is headed before it gets there. The goal is automated tooling that tracks these trajectories—without a human annotating every step.
Useful background: pretraining or post-training experience where you cared about what happened internally; continual learning methods (plasticity loss, catastrophic forgetting, task-boundary effects, etc.); or building automated interpretability pipelines (SAEs, activation patching, circuit discovery, etc.) rather than just running existing libraries. PyTorch or JAX, solid fundamentals, able to move fast.
Apply →We have an exciting opportunity to join the Oxford Martin AI Governance Initiative as a Postdoctoral Researcher in AI and Risk. The postholder will lead research on AI risk, governance, and the societal impacts of advanced AI systems, combining high-quality academic research with policy-facing activity and engagement with policymakers and international organisations.
Apply →The Technical AI Governance (TAIG) Programme is Oxford's dedicated funding vehicle for research at the intersection of technical AI safety and governance. TAIG offers funded studentships and fellowships for exceptional candidates—this is one of the few routes to fully funded doctoral and postdoctoral research in this space at Oxford.
If you are applying for a DPhil or postdoctoral position at TSG Lab, we strongly encourage you to explore TAIG funding—particularly if you are an overseas student concerned about fees.
Explore the TAIG Programme →TSG Lab accepts DPhil students through the following Oxford programmes. TAIG studentships can fund any of these routes.
For guidance on how to apply and what to include in your email, see our Contact page →