What you’ll do
- Replicate published agent-attack techniques and extend them against our reference stack.
- Build evaluation harnesses for agent guardrails and policy engines.
- Co-author a public write-up of your findings.
What we’re looking for
- Pursuing a degree in CS, security, or a related field.
- Prior project work in security CTFs, red-teaming, or applied ML safety.
- Strong writing skills — you'll publish.