Senior Technical AI Safety Research Manager

Senior Technical AI Safety Research Manager Help shape the future of AI safety research

We're looking for a strong research manager to support our AI safety programmes. This is a high-trust, technically-grounded role: you'll work directly with researchers to scope projects, provide mentorship, and find the best path to impact on AI risk. There’s also plenty of room to grow and take on new initiatives as Meridian evolves.

You’ll need technical depth, judgement on research directions, strong organisation, and a real commitment to creating impact and helping researchers flourish.

Meridian is a research hub and community for AI safety researchers based in Cambridge, UK. We host a strong community of full-time researchers, run structured research fellowships for early- and mid-career researchers, and convene the local AI safety community through workshops, reading groups, and events. We’ve incubated Geodesic Research, a new technical AI safety research lab focused on alignment priors before RL, and ERA, a world-class research fellowship.

Our own programmes – including MARS (Mentorship for Alignment Research Students), and our Research Accelerator Weeks (RAW) and Visiting Researchers Programme (VRP) – have supported over 300 researchers and contributed to NeurIPS, ICML, ICLR, and the UK government AI agenda.

• Scope and review research projects for MARS fellows and Visiting Researchers
• Provide technical feedback on proposals, drafts, and research plans
• Run regular check‑ins with researchers; identify and unblock obstacles
• Connect researchers with relevant people, papers, and opportunities
• Some minimal ops, ensuring that incoming VRs have a friendly point of contact and the resources they need
• Help to shape impactful research proposals during our Research Accelerator Weeks
• Choose our cohorts for RAW, and help review candidates for MARS
• Run red‑teaming and proposal development sessions, lab meetings and journal clubs

• A strong commitment to acting on the most serious risks from advanced AI
• Experience on a top fellowship, such as LASR, ERA, MATS, or Astra, or equivalent research experience in AI safety
• Publications in AI safety or machine learning
• Strong familiarity with the AI safety research landscape: current debates, agendas, who's working on what
• The judgement to distinguish promising research directions from dead ends, and solid ToCs from wishful thinking
• Strong empathy, and a desire to help other researchers flourish

• Experience leading, mentoring or supervising researchers
• Experience managing research programmes or collaborations across multiple institutions
• A network in the AI safety community
• Unlimited PTO
• Pension scheme with employer matching up to 4%
• Conference travel budget
• Daily lunch and snacks

• Location: In-person in Cambridge, UK
• Contract: Full-time (40 hours/week), initial 12-month contract; strong intention to renew
• Start date: Flexible; from summer 2026

• Monday: Check-ins with the current MARS cohort – one group is stuck on how to frame their AI control project; another needs help narrowing scope. Review weekly updates from visiting researchers. Skim a new paper on activation oracles that's getting attention.
• Tuesday: 1:1s with two VRP researchers. One is making good progress; the other is going down a rabbit hole that probably won't pan out – you help them see it before they've sunk another month. Office hours in the afternoon; a few people drop by with half-formed ideas they want to stress‑test.
• Wednesday: Run the weekly lab meeting – this week it's a visiting researcher from MATS presenting early results. Afterwards, a call with a mentor at Anthropic about a fellow who might be a good fit for their team.
• Thursday: Deep work day. You're reviewing three research proposals for the next MARS cohort and giving detailed written feedback on a draft that's almost ready for submission. Block out time to read a paper on evaluation awareness.
• Friday: Journal club on a recent ICML paper. Grab coffee with a biosecurity researcher who's thinking about pivoting to AI safety and wants your read on the landscape. Then a conversation with leadership about how Meridian's programmes are evolving and where research can have the most impact on AI risk.