Expert mathematicians are invited to author and verify high-quality open-ended prompts for AI model evaluation. In this role, you will craft and review challenging, unambiguous mathematical problems across core subdomains, assessing AI reasoning quality and helping establish rigorous evaluation standards for frontier language models.

You will be assigned one of two task types:

• Authoring Task: Create 5 original, open-ended prompts from your assigned subdomain at varying difficulty levels (undergraduate, advanced undergraduate, or graduate/professional). Prompts should require human judgment to evaluate the quality of the AI''s response, such as chain-of-thought reasoning or proof construction.
• Verification Task: Review 5 authored prompts for clarity, scope alignment, difficulty accuracy, and uniqueness. Edit prompts and difficulty ratings where needed.

Probability & Statistics, Algebra (including Linear Algebra), Ordinary/Partial Differential Equations & Dynamical Systems, Geometry, Graph Theory, Number Theory.

• Author clear, unambiguous, open-ended mathematical prompts that elicit evaluable AI responses.
• Verify prompts are within the scope of the assigned subdomain and correctly rated for difficulty.
• Ensure all 5 prompts in a task are sufficiently distinct from one another with varying difficulty levels.
• Apply expert judgment to assess the depth and quality of mathematical reasoning required.
• Edit prompts and difficulty assignments where standards are not met.

• Master''s degree or higher in Mathematics, Applied Mathematics, Statistics, or a closely related field.
• 2–6 years of professional or research experience in a quantitative field.
• Strong command of graduate-level mathematical concepts including proof writing, analysis, and formal reasoning.
• Experience in academic research, mathematical competition design, or quantitative industry roles is a plus.
• Excellent written English and ability to craft precise, well-scoped technical questions.

Expected commitment: 10+ hours/week. Asynchronous, fully remote work.