About the role:We're looking for a founding Research Engineer to study how AI agents behave in production. You'll join an early team, set the research agenda, and publish what we learn.
Your Focus:- Agent behavior: Discover how agents fail and respond to different conditions in the real world, and what that means for monitoring and safety.
- Simulations: Advance how we simulate agents and the situations they face before they reach users.
- Detection: Develop new methods for finding issues no one has defined yet.
- Open questions: High-upside bets on questions about agents we haven't thought to ask yet.
Your responsibilities:- Set Raindrop's research agenda and run the experiments behind it.
- Design evaluations and metrics for problems where the ground truth is unclear.
- Publish findings as papers, blog posts, and talks.
- Prototype new methods, and bring the ones that work into the product when it makes sense.
- Help define what research looks like at Raindrop as the team and product grows.
Ideal Candidate:- You make complex ideas clear through strong written and communication skills.
- Care about sharing your research through publications, open-source work, or public write-ups.
- Deep expertise in an area such as interpretability, evals, AI safety, anomaly detection, or representation learning.
- Good judgment about which experiments are worth running and when a result is real.
- You ship experiments and build prototypes
- Use and build AI agents and tools
