Astrophysics AI Problem Designer (Remote)

Mercor; Lacaussade; Sur place; EUR 83 000 - 119 000; Plein temps

Résumé du poste

Mercor seeks an Astrophysics & Cosmology Expert to design challenging computational problems for AI systems, using real scientific software, cosmological data, and state-of-the-art models to push the boundaries of evaluation.

You will work remotely 15–20 hours per week, developing multi-step workflows, refining problems through testing, and collaborating asynchronously with researchers across disciplines, contributing to robust benchmarks and reproducible experiments with Python and Astropy.

Qualifications

• Graduate-level training in a STEM field (MS, PhD, or equivalent research experience).
• Proven proficiency with scientific software libraries through publications or professional work.
• Strong Python skills for writing problem setups and solution validators.
• Comfortable in a Linux/terminal environment with remote compute sandboxes.
• Available for at least 15–20 hours per week.

Responsabilités

• Design challenging computational problems to test AI systems using real scientific software.
• Develop problems requiring strategic planning and multi-step workflows for AI to uncover hidden data.
• Refine problems through iterative testing against state-of-the-art AI models.
• Utilize astropy and related tools for cosmological calculations and data analysis.
• Work independently to enhance problem designs based on feedback.
• Collaborate asynchronously with teams to ensure problem difficulty is optimal.

Connaissances

Python

Linux

Scientific software

Formation

MS/PhD or equivalent

Outils

Astropy

Description du poste

Mercor seeks an Astrophysics & Cosmology Expert to design challenging computational problems for AI systems, using real scientific software, cosmological data, and state-of-the-art models to push the boundaries of evaluation.

You will work remotely 15–20 hours per week, developing multi-step workflows, refining problems through testing, and collaborating asynchronously with researchers across disciplines, contributing to robust benchmarks and reproducible experiments with Python and Astropy.