Help define how frontier AI models reason about mathematics.
1\. Overview
We are looking for senior mathematicians to improve how frontier AI models reason about real mathematical research.
Your mathematical judgment is the substance of this role. You will review the quality of mathematical reasoning tasks and model outputs, write the problem specifications and reference solutions and proofs that define what "correct" looks like, and design benchmarks that show whether a model is genuinely improving. We are looking for a practicing specialist rather than a generalist.
This role is fully remote.
2\. Key Responsibilities
- Expert review: Assess mathematical tasks and model outputs, catching gaps in proofs, unjustified steps, incorrect generalizations, and arguments that read well but would not survive peer review.
- Problem specs and reference solutions: Write precise problem statements and rigorous reference solutions and proofs that reflect how research mathematics is actually done.
- Benchmarks: Design challenging, research-level problems and evaluation sets in your area of specialization.
- Calibration: Work with other researchers and experts in adjacent fields to turn tacit expert judgment into explicit, teachable criteria.
3\. Core Qualifications
- Education: PhD in mathematics, applied mathematics, statistics, operations research, or a closely related field.
- Seniority: A faculty appointment, principal investigator, staff or senior scientist, or an equivalent senior research role at a university, national laboratory, or research organization.