Remote Math Competition Problem Writers, Medalists, Participants Affiliates - AI Trainer

Role Overview

Mercor is partnering with a leading frontier AI research lab on an exciting math project. We are seeking individuals with demonstrated expertise in Olympiad-style mathematics and problem setting. We’re inviting math competition problem writers, selection committee members, medalists & participants to help push the boundaries of AI reasoning by training large language models to tackle Olympiad-level problems better than the world’s top competitors.

This role is a short-term, high-impact research engagement, with the possibility of extension based on performance.

• Write original & advanced IMO-style mathematics problems to challenge frontier AI models
• Evaluate AI-generated solutions for correctness, clarity, and style
• Identify logical flaws, incomplete reasoning, or insufficiently-rigorous proofs
• Provide clear, concise, and correct solutions & chain of thought reasoning
• Maintain extremely high standards of mathematical precision and problem difficulty

Note:

Applicants must be highly proficient in writing and formatting documents using LaTeX

You are a good fit for the project if you have any of the following experiences :

• Problem writer
• Coach
• Selection committee member
• Medalist (gold, silver, bronze)
• Participant

And have participated in any of the following mathematics competitions (

Note:

Preference will be given for IMO & RMM participation, but we welcome applications from all competitions below):

• IMO
• IMO TST
• RMM
• USAMO
• APMO
• IMC
• EGMO

• Expertise across algebra, combinatorics, and number theory
• Ability to clearly articulate complex reasoning in written form
• Rigorous attention to detail in verifying solutions
• Independent, reliable, and disciplined work style

• Expected commitment: 20 hours/week minimum; up to 40 hours/week available
• Approximate project length: ~2 months, with extension potential
• Rolling start dates; setup typically within 1–2 days of approval

• This engagement uses a pay-per-task model, with payments tied to each fully completed and accepted problem or review.
• Compensation per task varies based on factors such as problem difficulty, solution quality, and overall efficiency, and contributors are free to choose how many tasks they complete.
• Based on current project rates, the estimated effective range for this engagement is approximately $54–$93 per hour of focused work, depending on the volume and quality of tasks completed. This range is an estimate, not a guarantee or cap on total payments for services rendered.
• There is no cap on total weekly payments for services rendered; contributors may complete as many available tasks as they choose to accept during the engagement.

Complete a brief interview (20–30 minutes)

Submit a short form detailing your experience writing problems and competing in math competitions