Mathematician Talent Network

$60–80/hr · Mercor · Part time

Apply your mathematician talent network expertise to evaluate and improve AI model reasoning through structured problem-solving and assessment.

What you would do

Research methodology

1.Deep expertise in core concepts
To verify you can apply fundamental knowledge to novel situations where textbook answers don't immediately apply.
For example: “Explain how you would approach a mathematician talent network problem where two standard approaches yield conflicting results.”
2.Judgment on ambiguous cases
Real work often involves edge cases without clear-cut answers; this tests whether you reason through uncertainty systematically.
For example: “Describe a time when you had to make a judgment call on a borderline case in your field and how you justified it.”
3.Communicating technical reasoning
Explaining your work to non-specialists is critical since this output trains AI systems that must reason across domains.
For example: “Walk through how you would explain a complex concept from your field to someone learning it for the first time.”
4.Error detection and pattern recognition
Volume reviewers develop intuitions for common mistakes that require explanation to teach an AI system.
For example: “Given 10 work samples from different reviewers, identify which patterns of errors recur and how you'd coach someone to avoid them.”
5.Research methodology and rigor
Validating that AI can carry out research-grade reasoning requires specialists who know what rigor looks like.
For example: “Describe the most challenging technical hurdle in your recent published work and how you validated your solution.”

We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.