About the role
The team builds and maintains Devin, an AI software engineer. The research engineer designs and executes evaluations and red-teaming exercises to ensure the agent's safety and correct behavior. This role also involves developing methods to align the AI's actions with desired outcomes and shaping its operational framework.
Our summary of the posting; the original is on the Cognition careers page.
What the posting requires
- Required
- Agent failure analysis
- Python
- PyTorch
- Eval infrastructure
- Experimentation at scale
- Training code
- JAX
- Inference code
- Agentic systems
- Tool misuse
- Preferred
Nothing listed as preferred.
These tags are our reading of the posting. They can miss something; the original posting is the reference.