Remote Physics Research Scientist for AI Benchmarks

Synthires; Greater London; Remote; GBP 82,000 - 138,000; Part time

Benefits offered by this job

Weekly payments

Fully remote work

Independent contractor engagement

Job summary

Synthires is seeking a Research Physics Expert for a fully remote, hourly engagement. The role involves 10 hours per week on advanced AI benchmarks, solving complex physics problems and creating reference solutions.

You will author Python-based templates, review submissions, and help develop high-quality benchmark datasets for AI systems. Ideal candidates hold a PhD or equivalent, with published research and strong Python, Jupyter, SymPy, and LaTeX skills.

Qualifications

• PhD or equivalent research-faculty level.
• Expertise in high energy, mathematical, biophysics, statistical, condensed matter, AMO, cosmology/astrophysics, quantum information or quantum mechanics.
• Strong publication record demonstrating research expertise.
• Experience solving complex research-level physics problems and developing rigorous analyses.
• Proficiency with Python, Jupyter Notebooks, SymPy, and LaTeX.

Responsibilities

• Solve research-level physics problems and develop complete reference solutions.
• Create structured problem decompositions and verification methodologies.
• Author Python-based answer templates and automated validation functions.
• Review and audit submitted solutions for correctness, rigor, and methodological soundness.
• Provide detailed feedback to improve solution quality and benchmark reliability.
• Adjudicate between competing solution approaches and determine final reference answers.
• Document derivations, verification procedures, error tolerances, and scientific rationale.
• Contribute to the development of high-quality benchmark datasets for evaluating frontier AI systems.

Skills

Python

Jupyter Notebooks

SymPy

LaTeX

Education

PhD or equivalent research-faculty level

Job description

Synthires is seeking a Research Physics Expert for a fully remote, hourly engagement. The role involves 10 hours per week on advanced AI benchmarks, solving complex physics problems and creating reference solutions.

You will author Python-based templates, review submissions, and help develop high-quality benchmark datasets for AI systems. Ideal candidates hold a PhD or equivalent, with published research and strong Python, Jupyter, SymPy, and LaTeX skills.