OpenAI
Mid
Scale compute-on-context research for OpenAI's frontier agents in Codex and ChatGPT
Interview for OpenAI's Agent Post-Training team, focused on scaling the compute spent on context as a lever for agent capability, with a concrete product surface in Codex Chronicle. Covers RL/post-training pipelines, evals and graders, and turning messy model-behavior failures into concrete experiments and fixes. Suited to candidates with hands-on LLM/RL/post-training or production ML systems experience who can move from a vague behavioral problem to a shipped fix.
Step into this interview
Free · a live voice mock calibrated to this exact role
What this interview tests
- Compute-on-context scaling experiments
- RL/post-training pipeline design (data, rewards, graders)
- Building evals/environments that expose model failures
- Debugging model behavior and turning it into training fixes
- Cross-functional translation of research into shipped product improvements
Common question themes
Design an experiment to test scaling compute spent on context
How would you build an eval or environment that surfaces a real agent failure mode
Walk through debugging a hard, messy model failure end to end
How do you decide a result is ready to go into a major training run
Balancing benchmark movement against real product/user impact
Related interviews
OpenAI
Senior
Forward Deployed Engineer
OpenAI
Senior
Data Scientist, Identity
OpenAI
Senior
AI Deployment Engineer
Cohere
Senior
Senior Software Engineer, Agent Infrastructure
Figma
Senior
Software Engineer, AI Platforms
Mid