Announcing the MATS Residency: A New Path for Experienced Researchers Working on AI Safety, or Moving Into it
Applications for the 2027 Winter intake open on August 25th, 2026 and close on October 31, 2026 (AoE)
MATS has accelerated over 630 AI safety researchers, who have coauthored 215+ papers with 17,000+ citations and founded organizations including Apollo Research and Resolution. We are now launching the MATS Residency.
What the Residency aims to do
The Residency provides researchers with 6 to 24 months (depending on the Residency track) of dedicated support to enable Residents in pursuing an ambitious research agenda focused on reducing catastrophic risks from AI. Addressing foundational challenges, including maintaining oversight of systems that outstrip human review, building evaluations and safety cases that make safety decidable, and making agentic and multi-agent systems safe, often requires years of sustained, concentrated effort. To meet this demand, the Residency is structured to allow Residents to take calculated risks without fear of unconventional outcomes. Every Resident leaves the program with a substantial body of reviewable work, four to eight weeks of post-Residency transition support, and continued access to the MATS research community.
What Residents will receive through the program
- Compensation: $155,000–$285,000 (USD, gross), depending on your Residency track, level of experience, and local cost of living at your office location
- Health: Comprehensive medical, dental, vision, and life insurance
- Retirement: Roth 401(k) and Traditional 401(k) plans for US-based employees, or a pension scheme for London-based employees
- Paid Time Off: 17 days of annual paid time off for US-based employees, or 20 days plus bank holidays for London-based employees
- Compute: Research compute with no preset cap, provisioned against research need
- Work Arrangement: Hybrid model with in-office presence required for at least one-third of the Residency
- Visa and relocation support: Visa support as needed, plus up to $25,000 in relocation assistance (covering eligible costs for a partner and dependent children)
- Location: MATS offices in London, Berkeley, or Washington, DC.
- Perks: Dedicated budget for attending conferences, workshops, and other research events, plus catered lunches and dinners for in-office workdays
- Community: Access to current Residents, Fellows, staff, and a network of mentors in safety organizations
Who the Residency is for
The Residency is for experienced researchers with a substantial body of work who want to do ambitious AI safety research with a high degree of independence. Whether you already work in AI safety or are moving into it from an adjacent field such as ML, policy, or governance, we encourage you to apply. We prioritize the quality and seriousness of your work over academic titles, credentials, or formal publications. Many of our strongest candidates include those whose best work is confidential, unpublished, or outside traditional research roles. If you have a serious body of work but lack conventional credentials or publications, we would rather you apply than rule yourself out.
The Residency is built for longer-term projects that do not fit neatly within conventional academic or industry settings and that benefit from sustained operational and research support at key decision points. That support is designed to strengthen your research while preserving your intellectual ownership and research independence.
Feel free to reach out with questions to residency@matsprogram.org.
How the Residency is structured
The Residency runs from 6 to 24 months and offers three parallel tracks (Bridge, Anchor, and Incubation) tailored to different career goals and research trajectories.
Bridge: The Bridge track is for experienced academics, researchers, and industry professionals from adjacent fields who already have research expertise but need a dedicated runway to build a credible AI safety portfolio. This includes ML professors, policy researchers, and senior industry researchers connecting their work to safety-relevant questions; MATS alumni temporarily unable to join a frontier lab; established academics considering a sabbatical or postdoc to redirect their research toward AI safety; and recent PhDs seeking an alternative to a conventional postdoctoral position.
This track is not designed for those who are new to research. It assumes an existing track record that needs redirecting toward AI safety rather than one that is being built from scratch.
Anchor: The Anchor track is for capable researchers who want a two-year staff researcher role rather than a PhD or postdoc, and need an institutional setting to develop an independent research agenda. It’s built for engineers, evaluators, and technical or policy practitioners whose work is fundamentally research, but don’t fit a traditional academic trajectory. This is not a fit for those seeking a conventional academic credential or short-term commitment; the Anchor track is a long-term, staff-level commitment.
Incubation: The Incubation track is for researchers founding new AI safety research agendas or organizations, who need structural support to move from idea to execution. Applicants may apply as an individual or a founding team can submit one application collectively. At least one applicant on a team must demonstrate safety-relevant research experience or evidence of research capability. It is not a fit for those seeking to join or extend an existing research agenda; the Incubation track is specifically for those who are ready to build a new research agenda, or the organization to pursue it, from the ground up.
You can read more about each track on our website.
Application requirements
To apply, you will submit an application form containing the following components:
- A short profile and eligibility section: Your contact details, current affiliation, CV (maximum 2 pages), office preference, availability for the intake, work-authorization status, highest degree, and any previous MATS involvement.
- Track choice: Your chosen track, and a two-to-three-sentence note on why it fits. Reviewers use this note to spot early on whether a different track might suit you better.
- References: Submit two references who can speak directly to the quality of your work. If your strongest work is confidential or was completed as part of a larger team, your referees may attest to your contribution in place of publicly available evidence. Some tracks have additional requirements for referees. Read more about track-specific requirements on our website.
- Representative work sample: Submit one work sample that best represents your strongest or most relevant work. This might include a paper, repository, technical report, evaluation, policy analysis, or another substantial research output.
- Written responses: Complete two short written responses and a brief counterfactual question about what you would do if you were not offered a place in the Residency.
Application timeline
The program accepts up to 30 Residents per year across three annual intake windows with rolling start dates.
August 25, 2026: Application window opens for the 2027 Winter intake
October 31, 2026 (AoE): Application window closes
Late November 2026: Initial review decisions sent to all applicants
First week of December 2026: Shortlisted candidates are invited to interview or complete a paid work test, depending on the track and research area they applied for
Late December 2026: Final offers extended to selected Residents
Early 2027: Winter intake begins
Upon acceptance, each Resident will collaborate with the MATS team to create an Individual Onboarding Plan outlining the Resident’s research block schedule, track-specific requirements, peer accountability partner (where applicable), support needs, office arrangements, and IRB status where relevant.
The next application window after this opens in March 2027.
How to apply
Applications are now open. Submissions are due by October 31, 2026 (AoE).
If you have any questions about the program or the application process, contact us at residency@matsprogram.org.