Research Manager, AI Safety

You support a group of fellows through the AI Safety Fellowship, so that each one does strong work and leaves with a clear next step. Salary: £86,750–£121,500 per year. We may be able to pay more for exceptional candidates. Location: London, at our office at LISA Hours: 0.5 to 1 full-time equivalent Focus: From technical AI safety to AI governance possible Visa: We can sponsor a Skilled Worker visa Why work at Pivotal Pivotal finds people who can work on the biggest risks from advanced AI and gets them working on them. Since 2024, 135 fellows have taken part in our fellowships, and alumni now work at METR, Apollo Research, UK AISI and GovAI. We have grown from 2 staff in 2024 to 10 today, and we plan to reach at least 16 in 2027. Read more about us at pivotal-research.org. What you will own You'll be the Research Manager for about 6 research fellows, supporting each project from scoping through to publication. You'll collaborate closely with our whole research management team. You'll engage deeply with your fellows' work: reading drafts, questioning experimental designs, suggesting framings, and helping projects stay ambitious but tractable. You’ll also hold the thread that keeps projects moving and fellows well supported. The research support includes: Run weekly 1:1 check-ins with each fellow; resolve bottlenecks quickly and adjust plans based on results. Give clear, actionable feedback on drafts, figures, methodology, and experimental design. Work with fellows and their mentors to scope projects that are both tractable and high-impact. Coordinate mentors effectively: identify gaps in mentorship coverage and fill them yourself or through targeted introductions. Help fellows with career planning, grant applications, and post-fellowship transitions. Collaborate with the rest of the Pivotal team on infrastructure, and the overall fellowship experience. Research management at Pivotal As an individual researcher, you might advance one or two projects. In this role, you'll have a hand in about 6 projects simultaneously (and about 18 projects every year). Your leverage comes from making each fellow meaningfully more effective than they would be on their own. The AI safety field still has a limited talent pipeline, and many of the people who will shape it over the next few years are early in their careers, still developing their research taste, and often short on high-quality guidance. This role sits at that bottleneck. You’ll help the next generation of researchers produce stronger work, find the right collaborators, and move into positions where they can have outsized impact. If you get energy from seeing other people's work click into place, this role may let you shape the field more than you could as a single researcher. Who does well in this role We don’t expect candidates to arrive as fully formed research managers. We do expect strong potential, good judgment, and real motivation to grow quickly in the role. The traits & backgrounds below are signals we think may matter. Research experience in AI safety, machine learning or AI governance, enough to judge whether a project is on track and to give useful feedback. Good research taste: you can tell which projects matter and which are likely to stall. Strong project management across many projects at once. Ability to keep momentum across multiple projects simultaneously — balancing attention to detail with the ability to context-switch. Care for people: fellows trust you and tell you early when things go wrong. Clear, direct communication, in writing and in person. A network in the AI safety field is a plus, because it helps with mentors and with fellows' careers. We don’t expect anyone to be strong in all of this. If the role excites you, please apply, even if you match only part of what we describe here. How we hire We review applications as they come in, so applying early helps. The process has these stages: Written application. A short form and your CV. It takes about 30 minutes. Automated video interview. About 10 minutes. You record short answers to a few questions, at a time that suits you. Work task. About 2 hours, based on real work in this role. We pay you for this time. You may use AI tools, as you would in the job. Tell us which ones you used and how. Personal interview. About 45 minutes. We discuss your work task and how you would approach the role. With your consent, we record it so that more of the team can review it. Work trial and references. Up to three paid days working with us at our office in London. We also speak to two referees, after asking you. If you need an adjustment at any stage, such as extra time, tell us and we will arrange it. If you have questions, email [email removed - click apply for more details].