Hey! alxk is a security and platform engineer with a decade in the field. He built the cybersecurity evaluation harness at Google DeepMind, which grew to run most of its dangerous capability evaluations. He also threat modelled AI deployments and worked on adversarial testing of image and text watermarking for generative models. He previously led the platform and security team at TrueLayer.

❏ Papers. Credited as a collaborator.

⚐ Talks.

♨ Tools and projects.