Astra is Constellation Institute’s full-time fellowship for empirical AI safety research and strategy and governance work. Fellows develop projects with senior mentors, weekly research management, and career support.
The primary location is Constellation’s Berkeley research center, with a London hub for UK-based fellows and rare remote participation. The September 2026 to February 2027 cohort is closed, and the site currently offers only an expression-of-interest route for a future cohort.
Benefits & funding
- Fellows receive an $8,400 monthly stipend for the five-month program.
- Empirical fellows may receive up to $15,000 per month in computing resources.
- Strategy and governance fellows receive a $2,000 budget for relevant workshops and conferences.
- The program includes weekly mentorship, weekly research-management meetings, career support, community access, incubation support, and visa guidance.
Eligibility
- Applicants must be able to work full-time on projects intended to reduce catastrophic risks from advanced AI.
- The empirical stream seeks strong Python skills, basic machine-learning and AI safety knowledge, initiative, and rapid learning.
- The strategy and governance stream seeks deep familiarity with catastrophic AI risks, clear communication, complex reasoning, and autonomous work; policy, technical, research, or grantmaking experience is preferred rather than required.
- The published pages state no nationality, residency, degree, age, gender, or years-of-experience gate; international applicants may receive visa support.
- Participation is primarily in person in Berkeley; a London hub is available to UK-based fellows, and remote arrangements are considered only rarely.
Can you apply?
Open to all applicants
No nationality, residency, education, or experience limits are listed.
Circlebreaker is a free database of global fellowships, scholarships and programs — every one decoded like this.
We only check the criteria we have data for. Always confirm full eligibility on the official page before applying.
Application process
- 1While applications are closed, register interest to receive notice of the next cohort.
- 2When applications open, submit a resume and at least two references; empirical applicants may add a CV, website, or writing samples, while strategy applicants are strongly encouraged to provide writing samples.
- 3After screening, empirical candidates complete a coding test, referral check, work tests, and an interview; strategy candidates complete mentor-specific screening, a referral check, and an interview.
Program timeline
Applications closed for the September 2026 cohort.
May 3, 2026
Program starts.
September 2026
Program ends.
February 2027
Date not announced
Next application cycle opens.