Remote Math Benchmark Architect for AI Research

Mercor; United States; Remote; USD 90,000 - 150,000; Full time

Job summary

Mercor seeks expert mathematicians to author and review high-quality academic assessment content for an AI research initiative. You will write and verify rigorous MCQs across core mathematics domains, evaluate solution quality, and help establish gold-standard benchmarks used to advance AI capabilities.

You will be assigned one of two task types: Question Authoring or Question Verification, rate difficulty, provide 1 correct answer with 9 plausible alternatives, and cite 1–5 academic references

Qualifications

• PhD or doctoral candidate in Mathematics or closely related field.
• Strong command of graduate-level mathematical concepts and formal proof writing.
• Excellent written English and ability to express complex ideas clearly.
• Experience with rigorous academic problem design or competition writing is a plus.

Responsibilities

• Author original math questions testing deep conceptual understanding.
• Rate each question's difficulty: Medium, Hard, or Expert.
• Provide 1 correct answer and 9 plausible distractors.
• Write step-by-step Chain-of-Thought solutions in Markdown format.
• Supply 1–5 academic references per question from reputable sources.
• For verification tasks: flag issues with clarity, completeness, precision, or solvability and justify edits.

Skills

Advanced mathematical knowledge

Proof writing

Academic English writing

Editorial review

Education

PhD in Mathematics or related field

Master's degree considered

Job description

Mercor seeks expert mathematicians to author and review high-quality academic assessment content for an AI research initiative. You will write and verify rigorous MCQs across core mathematics domains, evaluate solution quality, and help establish gold-standard benchmarks used to advance AI capabilities.

You will be assigned one of two task types: Question Authoring or Question Verification, rate difficulty, provide 1 correct answer with 9 plausible alternatives, and cite 1–5 academic references