Applied Mathematics Benchmark Specialist

What you'll do

You build benchmark questions that AI systems are tested on. Mercor assigns each contributor one of two task types:

• Question authoring: write original, hard multiple-choice questions in your area, rate their difficulty and submit them for review
• Question verification: check pre-written questions for accuracy, clarity and rigor, edit as needed, rate difficulty and document every change

The domains are specifically applied: signal processing, financial mathematics and actuarial science, mathematical economics, mathematical modeling of ecological and biological systems, mathematical programming and combinatorial optimization, and geomathematics and climate modeling. A pure algebraist or topologist will find less to work with here.

Each question needs 1 correct answer and 9 plausible but subtly wrong alternatives designed to trip up expert solvers, plus a step-by-step chain-of-thought solution in markdown and 1–5 academic references from reputable sources. Questions must be self-contained and test deep understanding rather than recall. Difficulty is rated as Medium (intro undergraduate), Hard (advanced undergraduate) or Expert (postgraduate and above).

Most of the time goes into the nine distractors. Each should come from a believable mistake (a sign error in a Fourier transform, the wrong discounting convention, a relaxed constraint) rather than a random number.

Who fits

• A PhD or current doctoral candidacy in mathematics, applied mathematics, statistics or a close field
• A master's may be considered with exceptional depth in one subdomain
• Strong graduate-level command and formal proof writing
• Academic problem design or competition-writing experience is a strong plus
• Excellent written English

What it pays

$61–77 per hour, Mercor's published band. Expected commitment is 10+ hours a week, asynchronous and fully remote. Payments are weekly on Stripe or Wise.

Worth knowing

Good:

• Very active: Mercor showed 96 hires on this listing this month
• Doctoral candidates qualify, not just PhD holders
• Low 10-hour weekly floor and fully asynchronous
• No residence restriction published
• Weekly payment on Stripe or Wise

Less good:

• Only the six applied domains listed are covered
• The 10-option format makes each question slow to write well
• The pay band is modest for postdoctoral-level authoring
• H-1B and STEM OPT holders are excluded
• Independent contractor: projects can be extended, shortened or ended early

About this listing

Posted by Mercor as an hourly remote contract, confirmed open on 24 September 2026. Pay, domains, question format and qualifications are the ad's own. See Mercor.