The work
You will help evaluate how AI models respond to safety-critical questions about nuclear science and engineering. The work involves creating scenarios, reviewing judgments, and explaining the technical reasons behind each safety decision.
You will also improve evaluation materials after receiving structured reviewer feedback.
- Write prompts and scenarios that test the boundary between safe and unsafe nuclear requests.
- Review and classify prompts created by other experts as safe or unsafe.
- Explain clearly why a request is acceptable or crosses a safety boundary.
- Revise evaluation work based on structured reviewer feedback.
What the role takes
This is remote, flexible, task-based contractor work completed on your own schedule. The role is open to candidates who are physically located in the listed countries while performing the work.
- Master’s degree or higher in nuclear science, nuclear engineering, or a closely related discipline.
- Clear, precise explanations of the technical reasoning behind safety judgments.
- Fluent in English.
- Open to candidates physically located in the United States, United Kingdom, Canada, Australia, or New Zealand.
- Helpful background includes reactor operations, nuclear safety, safeguards, national laboratory research, AI evaluation, red teaming, or data annotation.
How it works
Apply on OpenTrain with your resume, then complete the application on the hiring site.