Fellowships
Funded time for individual researchers who want to work on recursive improvement in depth.
What a fellowship is
A fellowship is a focused stretch of funded time on one research thread where AI improves AI. You stay where you are, keep your IP and publish under your own name. It is not a job, an internship or a degree program.
What fellows typically get
- A stipend for the length of the fellowship.
- Compute and API credits sized to the experiments.
- A mentor working on a related loop, and a small cohort of peers.
- Support for travel to present the work, and editorial help writing it up.
What we ask
- At least one public output: a paper, dataset, tool, benchmark result or replication.
- A clear statement of the loop your work serves, and an honest report on whether it turned.
- Open source by default. If your output is best kept closed, tell us why up front.
- Short, regular updates so others can build on the work early.
Who is a fit
Independent researchers, engineers between roles, late-stage PhD students, and builders with a loop they cannot get to in their day job. People in places without an established AI research scene are especially welcome. Staff at frontier labs already have the resources and are not eligible.
Length and timing
Fellowships run from a few months to a year, depending on the project. Shorter, project-scoped residencies are possible too. We take expressions of interest at any time.
Apply for a fellowship
Choose Fellowship application on the contact page. Tell us the loop you want to work on and what you have built so far.