✨ AI Insights & Summary
This is a unique opportunity for mathematicians to engage with cutting-edge AI by designing challenging math problems that test frontier AI models. As an 'expert author,' you'll leverage specialized tools like Z3 or SageMath, write Python reference solutions, and calibrate problem difficulty to understand how AI navigates complex scientific reasoning. This part-time, remote role offers a chance to deepen your expertise in mathematical software, gain insights into AI capabilities and limitations, and earn up to $45/hour equivalent, all while contributing to AI evaluation for major tech companies.
Expert Author - Mathematics Problems for AI Evaluation
Mindrift connects specialists with project-based AI opportunities for leading tech companies, focused on testing, evaluating, and improving AI systems. This is a project-based opportunity, not permanent employment.
What This Opportunity Involves
You will design mathematics problems to challenge a frontier AI model. Each problem must have an answer verifiable by code and require a specialized tool like Z3, cvc5, SageMath, Macaulay2, or others (NumPy or SymPy alone are insufficient). Problems run within a sealed Linux container with the tool pre-installed and a programmatic judge.
As an expert author, you will:
- Select an anchor tool and design a problem that hinges on its usage.
- Write a Python reference solution and supply input files if necessary.
- Determine the numerical answer and the acceptable margin of error for the model.
- Test the problem against the AI model in batches, tuning the difficulty until the agent succeeds in only a small number of attempts (10-30% pass rate).
- Submit tasks for review by a senior specialist in your subfield to ensure high quality.
This process offers deep insights into how AI models approach complex scientific problems, where they succeed, and where they struggle, while also enhancing your command of the anchor tool itself.
What We Look For
- Mathematicians with experience in Python, open to part-time, non-permanent projects.
- A degree in Mathematics (Pure or Applied) or a related field.
- 2+ years of research, applied, or teaching experience.
- Proficiency in Python for writing reference solutions.
- Fluency with, or a strong willingness to independently learn, at least one scriptable mathematical package: Z3, cvc5, Macaulay2, Singular, CasADi, IPOPT, SDPB, G6K, fpylll, or SageMath.
- Ability to design problems that genuinely require a specialized solver.
- Strong written English (C1+).
- Prior experience with the listed tools is not required if you are ready to learn quickly.
How It Works
Apply → Pass qualification(s) → Join a project → Complete tasks → Get paid.
Project Details
- Estimated Time Commitment: 10–20 hours per week during active project phases.
- Compensation: Up to $45 per hour equivalent, depending on experience and contribution pace.
Please submit your CV in English and indicate your level of English proficiency.