Computational Mathematics Researcher (Task Author)

What you'll do

You turn a real piece of computational work from your own research into an evaluation task for AI models. A strong task is something you have actually run and debugged: a numerical solver, an optimization routine, a simulation, a statistical model or an algorithm you implemented and validated.

Each task has a main problem, a few connected subproblems, a Python reference solution and tests that check the result. The difficulty should come from the mathematics itself, not from withheld information. You also specify the metrics a strong solution should be judged by, without publishing the exact passing thresholds, and revise the task based on structured review feedback until it is approved. You may optionally review other experts' tasks against a short rubric.

Accepted research areas include numerical analysis, numerical linear algebra, differential equations and dynamical systems, optimization, probability and stochastic processes, computational statistics, discrete mathematics, computational geometry, topology, number theory and inverse problems.

Who fits

• A research-track Bachelor's student (for example with an REU, thesis or lab role), a Master's or PhD student, a postdoc or a recent graduate in mathematics, applied mathematics, statistics or a related field
• Hands-on experience writing your own code, preferably Python, to implement or validate mathematical methods; familiarity with Python or ML alone is not enough
• Able to work as an independent contractor without visa sponsorship; students on F-1 visas, including OPT and CPT, are not eligible

Preferred: research in numerical methods, optimization, stochastic processes or computational topology or number theory; public code, a thesis or papers; and prior annotation or AI evaluation work.

The form asks for your institutional affiliation, degree program and year, a link to research code you have written, your work authorization, and a calculation from your research that could become a task. Plan the last one before you apply.

What it pays

Pay is $200 per approved task, with each taking about 2 hours, which works out to roughly $100 an hour; the AfterQuery board headlines $100 per hour. There is no cap on how many tasks you take on. Because pay is per approved task, tasks that need revision cost you extra time without extra pay. The ad does not publish payment frequency. It says all accepted experts undergo standard background and identity verification.

Worth knowing

Good:

• Open to students and recent graduates, not just PhDs
• Your own research becomes the task, so there is little to learn from scratch
• Flexible, task-based work with no fixed weekly schedule
• No cap on task volume

Less good:

• Pay only arrives for approved tasks, and the 2-hour estimate may not hold for you
• No visa sponsorship, and F-1 students are excluded
• Background and identity verification for accepted experts

About this listing

Read from the AfterQuery Experts board on 18 October 2026, where it was posted on 16 October 2026. The per-task pay, eligibility rules and form fields are AfterQuery's own text. See AfterQuery.