Flexible remote fellowship to reduce AI risk
Connecting rising talent with expertise to tackle risks from AI
SPAR is a three-month, flexible fellowship where experienced professionals and talented students work on AI safety, policy and fieldbuilding projects with professionals in the field.
Express interest in the next round Browse projects
Spring 2027 applications open December 16 and close January 13.
New in Spring 2027
SPAR now also hosts non-research projects
Starting this round, we’re also accepting more hands-on projects in areas like communications, fieldbuilding, and AI policy.
Some of our mentors from Fall 2026
See them all-
Belinda Li Anthropic
-
Cody Wild Google DeepMind
-
Jess Bergs UK AISI
-
Rishi Bommasani Stanford University
-
Jeff Sebo New York University
-
Noam Kolt Hebrew University
-
David Bau Northeastern University
-
Ann-Kathrin Dombrowski FAR.AI
-
Bob Fischer Texas State University
-
Ivy Gilbert New York University Center for Mind
-
Toni Sims NYU Center for Mind
-
Lionel Levine Cornell University
-
Rauno Arike Aether Research
-
A. Feder Cooper Yale University
-
Jamie Hayes Google DeepMind
-
Pierre-Luc St-Charles LawZero
-
Alec Harris Pivotal Research Extension
-
Rebecca Baron Kairos
-
Alex Mark Cambridge Boston Alignment Initiative
-
Gary Abel Fourth Eon Biosecurity Institute
-
Isha Harris Fourth Eon Biosecurity Institute
-
Jack Kengott SaferAI
-
Anemone Franz University of Notre Dame
-
Noah Siegel Google DeepMind
-
Konstantinos Voudouris UK AISI
-
Maxime Riché Center on Long-Term Risk
-
Joshua Jacob Sampura Research
-
Rishub Jain Sampura Research
-
Thomas Jiralerspong Bengio Lab
-
Helena Tran Constellation
-
James Lester Generator Residency
-
Jonathan Richens Google Deepmind
-
Kevin Kohler Simon Institute
-
Adam Oberman McGill
-
Mohamed Amine Merzouk Mila - Quebec AI Institute & McGill University
-
Aditya Singh Anthropic Fellows
-
Andy Han NYU
-
Ariana Azarbal Anthropic Fellows Program
-
Darion Cassel Independent
-
Michelle Nie Center for a New American Security
What is SPAR?
A remote program pairing fellows with professionals addressing risks from AI. Fellows gain experience and guidance, and mentors get capable collaborators.
-
5–40 hours a week
Designed to fit alongside other commitments.
-
3 months
Structured work that ends in Demo Day, with posters, talks and a career fair.
-
Any experience level
Open to undergraduate students and experienced professionals with relevant background.
Selected Outputs Browse the Research Library
SPAR projects have been accepted at ICML and NeurIPS, covered by TIME, and led to full-time roles for fellows.
- ICML 2026 AI Security
PersistBench: When Should Long-Term Memories Be Forgotten by LLMs?
Mentored by Vyas Raina (University of Cambridge) and Ivaxi Sheth (CISPA) - ICLR 2026 Interpretability
Transformers Don't Need LayerNorm at Inference Time
Mentored by Stefan Heimersheim (Apollo Research) - Covered by TIME AI Governance
Lead, Own, Share: Sovereign Wealth Funds for Transformative AI
Mentored by Deric Cheng and Justin Bullock (Convergence Analysis) - ICML 2026 Chain of Thought
A Positive Case for Faithfulness: LLM Self-Explanations Help Predict Model Behavior
Mentored by Noah Y. Siegel (Google DeepMind, University College London)
Why apply?
-
Hands-on experience
Build skills by working on technical AI safety, AI policy, biosecurity, strategy, communications or fieldbuilding projects. SPAR and your mentor provide structured support along the way.
-
Expert mentorship
Work closely with experienced people in AI safety, who help you build the skills and judgment to work on reducing risks from advanced AI.
-
Career opportunities
Many fellows go on to full-time roles in AI safety. Through your project and the wider SPAR community, you can connect with people at organizations like METR, Redwood Research and GovAI.
-
Community
Join a team of peers with shared interests who support each other's growth. Virtual coffee chats and in-person meetups provide additional opportunities to connect.
-
Impact
Do cutting-edge work to help the world answer the most pressing questions of our time. Many projects lead to published papers or preprints with fellow co-authors.
-
Prizes!
Last year we awarded $7,000 total at our Demo Day, which rewards the best posters and talks for projects in the program.
What's it like to be a SPAR fellow?
From application to publication, here's what you can expect.
-
Browse Projects
Explore available projects and apply to work on ones you're excited about.
-
Work on your project
Spend 5-40 hours/week with your team on projects tackling large-scale risks from advanced AI. Receive structured support and regular feedback.
-
Present at Demo Day
Present your work at our virtual mini-conference attended by leading AI safety organizations. The best posters and lightning talks will be eligible for prizes!
-
Continue & Collaborate
Many mentorship pairs continue working together after the program ends, leading to publications and career opportunities.
- Rishub Jain CEO of Sampura Research, Tailwind grantee, ex-Google DeepMind
“Without SPAR, Sampura Research would not exist. SPAR was the only fellowship that accepted me as a mentor at the time, and the only one that allowed for large well-funded group projects. It was the most fulfilling work I had done that year, and revealed how talented and motivated researchers are outside labs, directly leading me to leave Google DeepMind and start an org. We have immediately hired 3 past SPAR mentees, dramatically speeding up what we could accomplish early in our org.”
- Andrew Draganov Research Lead, Arcadia Impact
“Mentoring with SPAR in spring 2026 gave me the chance to work closely with a small group of ambitious people. After the project I encouraged two of the Fellows to apply for our open roles at the alignment team at Arcadia Impact. Both now work full time on the team. As a way to get to know highly promising candidates, SPAR proved to be a successful path for our team.”
Fellow testimonials
“Thank you for such a fantastic opportunity! During the program I was offered an AI Safety role at Google and recently started right as SPAR ended. I felt like the program gave me great experience and helped strengthen my passion which no doubt helped with getting the job.”
“[My mentor] was extremely willing to share his connections and was involved in the trenches with us from the beginning. He provided some thought-provoking insights and guidance and opened himself up for 1:1s, even as he started a new key role... It was a pleasure to work with [my mentor] and I hope it won't be the last time.”
“I am currently working as a machine learning engineer in frontier ML ... SPAR gave me a clearer sense of what AI safety research looks like and helped me see that this is the direction I want to focus on next.”
“It's exactly the "foot in the door" one needs when transitioning into research positions.”
Am I eligible?
The program is open to undergraduate, graduate/PhD students, and professionals of different experience levels.
Mentors often look for a technical background (e.g., ML, CS, cybersecurity, math, biology, physics, etc.), knowledge relevant to policy and governance (e.g., law, international relations, public policy, political science, economics, etc.), or experience in communications, operations or community building.
Some projects may require additional background knowledge, like knowledge of specific research areas, techniques, or tools.
Even if you do not entirely meet these criteria, we encourage you to apply!
Many past fellows have been accepted even though they didn't completely match a project's criteria.
FAQs
What is the program timeline?
The Spring 2027 round follows this timeline:
- November 11 to 30: Mentor and organization applications
- December 10: Mentor decisions released
- December 16 to January 13: Fellow applications
- February 1: Fellow decisions released
- February 8 to May 10: Project period
- May 15: Demo Day
- After May 10: Optional extension period
What kinds of projects does SPAR host?
SPAR hosts projects across technical AI safety, AI policy and strategy, communications and fieldbuilding, grouped into streams. In Fall 2026, nearly 30% of projects were in AI policy, governance or strategy. Each stream has its own default selection tests, so what you’re asked to do depends on the projects you apply to.
Is there an age or education requirement?
Almost none. SPAR is open to anyone 13 or older, including high school students, and there's no degree requirement. Mentors select whoever can contribute best to their project, so relevant skills and motivation matter much more than your stage of education. Do be aware that the program is highly competitive, and earlier-career applicants are weighed against people with significantly more experience.
If you're under 18, please let us know in the “Anything else?” question of your application. We may need a parent or legal guardian to sign a consent form before you can participate.
Can I take part from anywhere in the world?
Yes. SPAR is fully virtual and our mentees join from all over the world. There's no country or residency requirement. The one caveat is that, as a US organization, we're required to comply with US sanctions laws, which may affect our ability to accept participants from certain sanctioned countries or regions.
If you live in a country under comprehensive sanctions, or you're affected by narrower sanctions in some other way (the US Treasury keeps a list of active sanctions programs), please write to us at spar@kairos-project.org (we're happy to confirm your eligibility before you apply), or flag it in the “Anything else?” question of your application.
Flagging it won't count against you in selection. We run sanctions screening on all participants during project matching anyway, so telling us early just means we can sort out your eligibility before you and a mentor invest in a match. If you know about a sanctions issue and don't tell us, we may have to withdraw you mid-program, which is far worse for everyone than sorting it out up front.
The other thing to check is time zones. Some projects ask for a weekly meeting in a particular window, and any such preference is listed on the project page.
Is participation in SPAR compensated?
We cover approved project expenses, including compute and access to APIs and LLMs. Some organizations that host projects plan to offer their fellows a stipend, as indicated in project pages. We also often fund ways to connect with others in AI safety, such as meetups with fellows in your area and travel to present your SPAR work at conferences.
SPAR is a volunteer research program. Participants don't receive wages, salary, or other compensation, and taking part doesn't create an employment relationship with Kairos or with your mentor.
If you're studying in the US on a visa (F-1, J-1, or similar), we recommend checking with your international student office before committing. If it helps, we can provide a letter confirming that SPAR is an unpaid volunteer program. Just write to us at spar@kairos-project.org.
What does SPAR expect of accepted applicants?
We expect accepted mentees to commit the hours they agreed to (typically 5-40 hours per week), attend regular meetings with their mentor and team, complete assigned research tasks, and present their work at the end-of-program Demo Day. We also expect professional communication and collaboration throughout the program.
Can I apply to multiple projects?
Yes, you can apply to as many as you'd like! However, it's rare to be matched with more than one as a participant.
You don't have to decide everything upfront, either. Until applications close, you can edit your submitted application and add more project applications at any time.
Can I edit my application after submitting?
While applications are open, your confirmation email includes a personal edit link for updating your answers, replacing your resume, and adding or removing project applications. Editing for the current round has closed.
How do I know my application went through?
You'll receive a confirmation email summarizing your answers for each application you submit. If you don't see it after a few minutes, check your spam folder before writing in.
Can I use AI to help write my application?
No. Every free-response answer must be written by you, without AI writing, rewriting, or polishing (including Grammarly-style tools and AI translation). Your answers are how mentors understand how you think, and we can and will reject applicants for AI-written answers. You're not graded on prose, so short, rough answers in imperfect English are completely fine. You can still use AI to figure out which projects to apply to, or for high-level feedback on your application.
There is one exception. If a disability makes it impossible or impractical to type your answers yourself and you write with assistive technology (for example, dictating and using AI to clean up the transcript), write to us at spar@kairos-project.org and we'll exempt you from this policy. Please also mention it in the “Anything else?” question of your application so our reviewers have the context. The exemption covers getting your words into text, and the ideas in your answers must still be entirely your own.
What if I miss the application deadline?
Applications close at the end of the date shown in the timeline above, Anywhere on Earth. In practice that means you have until noon UTC the following day, so nobody is cut off early by their time zone.
We can't grant individual extensions or accept applications after the round closes. If you miss this round, the next one starts about six months later, and we'd love to see you apply then. Fill out the expression of interest form and we'll email you when applications open.
Can I propose my own project idea?
Mentees apply to projects that mentors have already proposed, so you don't submit your own project as a mentee. You can still bring your idea, though. Apply to the projects closest to it and make the case in your application, and raise it with the mentor if they get in touch. Project scopes are often flexible, and mentors regularly shape the work around what their mentees are excited about.
If you want to run your own project outright, that's the mentor side of SPAR. Mentor applications open a few weeks before mentee applications, around November for the spring round and around July for the fall round.
How does the application and project matching process work?
Apply to as many projects as you're excited about. Mentors will review your application and may ask you to interview, complete a short assessment or trial task, or provide references (though not every project does this, so don't panic if you don't get an invite!). Then mentors will submit a ranked shortlist of project applicants, and you'll submit a ranked list of projects you've applied to. We'll run an algorithm that matches mentees with projects based on both mentor and mentee preferences, as well as other factors like ensuring no one gets assigned more hours of work than they're available for.
Please note that this means that if you're shortlisted for multiple projects, you won't get to choose between them, though you are more likely to be assigned to the ones you're more excited about.
Will I be asked to provide references?
Possibly. After applications close, some applicants will be asked to provide one or two references (people who can speak to your work or research, like a professor, manager, or collaborator) as part of the selection process. Only a subset of applicants will get a request, and not receiving one doesn't mean your application was rejected.
Referees can take several days to agree and respond, so it's worth thinking ahead about who you'd ask. You don't need to contact anyone before you receive a request.
When will I hear back about my application?
Applicants should expect to hear back on February 1. Decisions arrive by email. If you haven't seen anything by the following morning, check your spam folder, then write to us at spar@kairos-project.org.
Mentors may also reach out earlier to schedule interviews, ask clarifying questions, or request that you complete assessments or provide references as part of the selection process. Each request states its own deadline.
When is the next round? Is there a summer round?
SPAR runs twice a year. The spring round opens applications around December and the fall round around July. There's currently no summer round. The best way to hear when applications open is the newsletter at the bottom of this page.
There are lots of AI safety fellowship programs! Is SPAR my best fit?
SPAR is primarily designed for people who want to work on AI risk around other commitments. It's virtual and can be done part-time or full-time, so you can participate from anywhere. We also accept fellows with a wide range of experience levels, from students looking for a first project to experienced professionals looking for a fast way into the field.
Many people participate in multiple AI safety upskilling programs over time. 80,000 Hours maintains a list of these programs here. If you're interested in SPAR, you may be especially interested in the following (note this list may be out of date):
Introductory courses:
ML upskilling:
Policy mentorship:
Technical safety mentorship:
Hear when applications open for the next round
Fill out the expression of interest form and we’ll email you when applications open.