The opportunity
The ChatGPT Model Flywheel team unified goal is to transform model advancements into great ChatGPT user experiences through reliable serving, rapid experimentation, safe deployment, and continuous improvement.
What you'll do
Model Experimentation: Enable rapid, safe model validation for ChatGPT and Codex products through experiment automation and lifecycle management.
Model Deployment: Ensure safe, scalable deployment of model capabilities with robust rollout and operational tooling.
Automate capacity management and incorporate platform-wide health monitors.
Model Measurement: Build comprehensive evaluation and measurement systems for model quality, from user signals to launch scorecards.
Improve end-to-end feedback loops for continual model improvement.
Elevate and consolidate ChatGPT’s harness, context management, and system prompt frameworks.
What they're looking for
- Drive expansion and improvement of multi-tier model experiences.
- Support and scale self-serve experiment capabilities and automated guardrails.
- Lead model rollout automation, capacity management, and health monitoring.
- Shape end-to-end measurement systems (evals, grader signals, user feedback, etc.).
