Scrum Master Expert for Agile AI Evaluation

You will create and review evaluation data that tests whether AI models understand practical Scrum and agile delivery. Your work will turn real experience with facilitation and team dynamics into realistic scenarios, reference answers, and quality decisions.

• Write scenarios covering sprint planning, delivery challenges, and retrospectives.
• Create reference answers that explain the facilitation reasoning behind each intervention.
• Evaluate AI responses for agile soundness and awareness of team dynamics.
• Flag advice that sounds correct in theory but could harm a real software team.
• Apply practical judgment to backlog management, impediment removal, and team coaching.

What it pays and takes

This is fully remote, asynchronous independent contractor work. The engagement is ongoing, with a minimum commitment of 10 hours per week and flexibility to change your workload from week to week.

• $70 to $110 per hour, paid weekly through Stripe.
• Open to candidates in countries across North America, Europe, Asia, Africa, the Middle East, the Caribbean, and Oceania.
• Strong written English and the ability to explain facilitation reasoning clearly.
• At least two years leading agile delivery as a Scrum Master or agile delivery lead for software teams.
• Experience facilitating the full ceremony cycle, from planning through retrospectives.
• Practical knowledge of backlog management, impediment removal, and team coaching.
• CSM, PSM, or SAFe certification is helpful.
• Experience coaching multiple teams, scaling agile practices, or working as a developer, tester, or product owner is helpful.

How it works

OpenTrain is a platform where people find AI training and data-labeling work and start their applications. 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 AI systems, including writing scenarios and reviewing model responses. People with practical professional experience are paid to show models how sound decisions, clear reasoning, and useful answers should look.