Remote Math Benchmark Lead (Contract)

Crossing Hurdles; Colombie; À distance; COP 65 513 000 - 122 290 000; Plein temps

Résumé du poste

Crossing Hurdles is seeking a Pod lead for Math (Terminal Bench Science) in a remote contract role. You will review trainer-created tasks for mathematical correctness, verify convergence and error handling, and ensure tasks foster multi-step reasoning and robust computation.

You will design graders based on mathematical properties, provide feedback, and mentor trainers in task design and coding practices while collaborating with stakeholders to improve benchmark quality.

Qualifications

• PhD, postdoc, or equivalent in mathematics or related field.
• Programming in Python, Julia, C/C++, MATLAB/Octave.
• Linux proficiency and terminal usage for scientific computing.
• Experience with numerical methods, optimization, and mathematical modeling.
• Ability to evaluate algorithms, proofs, and computational results rigorously.
• Strong communication skills to provide constructive feedback.
• Experience mentoring or leading technical or research teams.

Responsabilités

• Review trainer-created tasks for mathematical correctness and reproducibility.
• Verify the convergence, stability, and error handling of mathematical tasks.
• Ensure tasks promote genuine multi-step reasoning and appropriate computation.
• Design graders based on mathematical properties and check for optimality certificates.
• Provide feedback and track trainer revisions to ensure quality.
• Mentor and onboard trainers in task design and coding practices.
• Collaborate with stakeholders to enhance benchmark quality.

Connaissances

Python

Julia

C/C++

MATLAB/Octave

Linux

Numerical methods

Formation

PhD or postdoc in mathematics

Description du poste

Crossing Hurdles is seeking a Pod lead for Math (Terminal Bench Science) in a remote contract role. You will review trainer-created tasks for mathematical correctness, verify convergence and error handling, and ensure tasks foster multi-step reasoning and robust computation.

You will design graders based on mathematical properties, provide feedback, and mentor trainers in task design and coding practices while collaborating with stakeholders to improve benchmark quality.