You will create and assess safety-focused evaluation tasks for AI models. The work involves deciding whether medical and radiation-related requests are safe, explaining each decision, and helping improve model behavior on safety-critical questions.
- Write prompts and scenarios that test the boundary between safe and unsafe requests in radiology, medical physics, or radiation science.
- Review and classify prompts created by other experts.
- Explain clearly why a request is acceptable or where it crosses a safety boundary.
- Revise evaluations in response to structured reviewer feedback.
What it pays and takes
This is remote, flexible, task-based contractor work.
- Pay: $85 per approved task.
- Location: Open to candidates in Australia, Canada, the United Kingdom, New Zealand, and the United States.
- Schedule: Complete tasks on your own schedule; no fixed weekly hours are provided.
- Education: Master's degree or higher, such as an MD, DO, MSc, or PhD, in radiology, medical physics, radiation science, or a closely related discipline.
- Language: Fluent in English.
- Required skill: Explain the technical reasoning behind safety judgments clearly and precisely.
- Helpful background: AI evaluation, red teaming, or data annotation experience.
- Relevant additional experience: Board certification in radiology or medical physics, or clinical or research work with radiation sources.
How it works
Apply on OpenTrain with your resume, then complete the application on the hiring site.
About AI training work
AI training is the human work behind systems that learn from examples, including writing and rating model responses and reviewing safety-related outputs. OpenTrain is a free platform where people find AI training roles, build a profile, and apply to projects across the industry.