You will design and implement challenge protocols to verify datacenter workload occupancy and measure accelerator performance under adversarial conditions. The role involves conducting research, publishing findings, and collaborating with policy and security teams to ensure trustworthy AI infrastructure.
You will design and build systems to verify AI datacenter computations using confidential computing and hardware attestation. This involves creating registries for third-party auditing and ensuring the integrity of attested pipelines across organizational boundaries.
The Head of Operations will lead the review and upgrade of the organization's operating stack, including hiring, payroll, and office management. They will also partner with leadership to refine organizational strategy, culture, and governance structures while overseeing scaling efforts across multiple jurisdictions.
The Lead Research Scientist will set and evolve the research agenda, identifying critical questions in AI safety and building the research programs to answer them. They will lead major research projects, mentor staff, and build collaborations with external AI safety institutes and model developers.
Design, implement, and run evaluations of misalignment and loss-of-control risks in frontier models. Build realistic agent environments, scaffolds, and tool integrations to study model behavior under increasing autonomy.
Own research projects end-to-end, from identifying important questions and formulating hypotheses to designing experiments and publishing results. Develop rigorous evaluations for misalignment and loss-of-control risks in frontier AI models while collaborating with research engineers.
You will build technical prototypes for AI verification and collaborate with external experts to translate research into practical policy tools. The role involves working across machine learning engineering, cybersecurity, and systems architecture to develop robust and auditable verification mechanisms.
