Postdoctoral research position in AI safety focusing on latent probing of large language models, activation monitoring, and conformal/inference-based robustness to detect misalignment and harmful behavior. 35 hours per week, fixed-term to 31 Oct 2027; PhD required or near completion; starting as...