Fellowships
Apply Now!Spring 2026 applications are now open! Apply by January 25, 2026 (end of day, anywhere on earth).
Our fellowships are 8-week, in-person reading groups that meet for 1.5hrs/wk. We offer two ways to engage:
- Self-contained with no outside readings nor capstone required.
- ~1 hour of required pre-reading each week plus a capstone project (e.g., a policy brief/memo, implementing an ML paper, proposing a research project, etc.).
The fellowships are open to undergraduates, graduate students, and postdocs. Catered lunch/dinner from local restaurants (not pizza!) is provided. Please direct questions to contact@tassa.dev .
Our technical fellowship covers the foundations of frontier AI technology. Topics include model interpretability, goal misgeneralization, reinforcement learning, oversight, red-teaming, and more.
Sample curriculum:
- Week 1: Introduction to modern machine learning and AI
- Week 2: AI risks and alignment challenges
- Week 3: Scalable oversight and adversarial training
- Week 4: Interpretability and model internals
- Week 5: Red-teaming and constitutional AI
- Week 6: Technical governance
- Week 7: Social impacts and civic AI
- Week 8: Field review and career opportunities
Note: We update and iterate our curricula each semester to reflect the latest developments in AI safety research.