Competitive Mathematics AI Training Problem Reviewer - United States

About the opportunity

Help improve advanced AI training work by reviewing competitive mathematics problems and research environments. This role focuses on determining whether problems are mathematically realistic, well designed, and correctly graded. You will evaluate the quality of each problem and its evaluation environment rather than solve the problem yourself. Your review will help create reliable, rigorous training material for frontier AI systems.

What you will do

• Review competitive mathematics training problems and related research environments.

• Assess mathematical realism, problem design quality, completeness, and grading correctness.

• Identify ambiguous instructions, flawed assumptions, incorrect grading logic, and other quality issues.

• Provide constructive feedback, detailed annotations, and clear recommendations.

• Apply rigorous mathematical judgment while following structured review guidelines.

• Maintain consistency across reviews by applying evaluation criteria carefully and objectively.

• Communicate findings clearly and work independently in a remote environment.

What we are looking for

• A master's degree or PhD in Mathematics.

• At least three years of relevant professional, academic, research, teaching, competition, or problem-review experience.

• Strong expertise in competitive mathematics and advanced mathematical reasoning.

• Experience reviewing, evaluating, grading, annotating, or quality-checking challenging mathematics work.

• Ability to identify subtle design or grading problems without treating the assignment as a problem-solving exercise.

• Strong structured written communication and the ability to provide constructive feedback.

• Careful attention to detail when checking complex mathematical material.

• Self-direction and reliability in a short-term remote assignment.

• A desktop or laptop with reliable internet access.

Engagement details

• Short-term remote contractor assignment with an immediate start.

• Approximately 15 to 20 hours of work.

• At least 4 hours of working-time overlap with Pacific Time.

• A background verification process that may include providing documentable information related to professional credentials and experience may be required as part of the onboarding process.

Compensation

• Actual compensation is $70 USD per completed task, not per worked hour.

• Each task typically takes 30 to 90 minutes.