Job Description:

We are seeking a Mathematics and Scientific Computing SME to develop and review advanced scientific coding problems for AI evaluation. You will turn research-level mathematical methods into reliable Python problems, reference solutions, and tests.

Key responsibilities :

- Create self-contained scientific coding problems involving numerical reasoning and implementation.

- Write correct Python reference solutions using appropriate scientific computing methods.

- Develop and debug tests, including standard, edge, ill-conditioned, and failure cases.

- Independently verify expected results and establish justified numerical tolerances.

- Review problem specifications, formulas, assumptions, code, tests, and evaluation results.

- Identify whether issues originate in the problem statement, solution, tests, expected results, or evaluation setup.

- Check convergence behaviour and ensure tests allow mathematically valid alternative implementations.

- Document required corrections and rerun corrected solutions and tests before recommending approval.

Required qualifications :

- MSc or PhD in Mathematics, Applied Mathematics, Computational Science, Engineering Mathematics, Financial Mathematics, or a closely related discipline.

- Advanced expertise in at least one area: numerical linear algebra, computational mechanics, computational finance, or a closely related field.

- Strong hands-on Python programming experience. Python is required for implementing and testing solutions.

- Ability to translate mathematical methods into correct, verifiable Python code.

- Experience critically reviewing technical or scientific code and validating numerical results.

Preferred qualifications :

- Experience with NumPy and SciPy.

- Knowledge of iterative solvers, numerical optimization, ODE/PDE methods, finite-element methods, numerical simulation, stochastic methods, or quantitative finance.

- Experience reviewing research code, numerical software, scientific benchmarks, or computational datasets.

- Familiarity with scientific coding benchmarks is helpful but not required.