Vals AI Fellowship 2026 (up to $2,500 stipend) – YOA

Deadline: June 30, 2026

Applications are open for the Vals AI Fellowship 2026. Vals builds evaluations for AI in authorized, finance, science, and different excessive-stakes domains. They work with the main AI labs and Fortune 500 firms, which suggests fellows get entry to issues and infrastructure which can be troublesome to replicate in a college lab. Their objective is to develop higher benchmarks, analysis strategies, and guarantee they’re measuring what matters- as a fellow, you’ll be serving to sort out a few of these issues!

Fellows apply with a proposal for a brand new benchmark they need to construct. If accepted, the fellowship is time and assist to design, implement, and validate that benchmark. Some domains they’re embrace:

• Long horizon agentic benchmarking in laptop use or software program engineering
• Cybersecurity
• Finance, legislation, insurance coverage
• AI for Science evals- analysis-degree arithmetic, biology, supplies science, theoretical physics, and extra.

Benefits

• Stipend: $1,000-2,500 per week
• Compute: Unlimited API credit + funds capability for GPUs and human knowledge
• Mentorship: weekly 1:1s with a Vals analysis lead, plus common entry to the broader staff
• Access: frontier mannequin APIs, their inner analysis infrastructure, and (the place applicable) actual buyer analysis issues
• Workspace: desk within the San Francisco workplace (when you’re keen on being in-individual)
• Network: intros to researchers throughout frontier labs

Eligibility

• Applicants ought to have most well-liked background in CS, ML, statistics, or an adjoining related discipline, though all purposes may also be thought of on the benefit of the concepts and staff.
• Genuine curiosity in analysis as a analysis self-discipline, not simply as a stepping stone.
• Comfortable working in a startup surroundings: sooner iteration, much less hand-holding, extra direct contact with work that finally ends up mattering.

You don’t want to have printed in ML venues particularly. Great analysis work can come from measurement, psychometrics, statistics, HCI, and social science backgrounds. They ask fellows to give you the option to commit no less than 20 hours per week throughout the fellowship. This may be completed throughout a depart, over the summer season, or throughout a versatile interval of a PhD program.

Application

Click here to apply

For extra info, go to Vals Fellowship.