AI Trainer (Physical Sciences)

You will create realistic computational physical-science tasks for AI models and define what a correct solution must include. You will also assess whether generated technical work is accurate, reproducible, and physically sound.

The work combines physical-science expertise with Python and computational research. You will give clear, technically grounded judgments on model outputs.

• Create task instructions that test calculations, code, simulations, models, and technical analysis.
• Set up reproducible environments, reference solutions, evaluation criteria, rubrics, and graders.
• Review AI-generated Python code, calculations, simulations, models, and other technical deliverables.
• Find invalid assumptions, numerical errors, physically impossible results, incorrect methods, and reproducibility failures.
• Set objective standards for correct solutions and explain your technical judgments.

This is expert-level contract work for people with prior paid experience training, evaluating, benchmarking, or improving AI systems on scientific or technical tasks.

• Time: Less than 20 hours per week.
• Language: Professional fluency in English.
• Experience: Expert-level experience with AI training or model evaluation involving scientific or technical content.
• Python: Strong practical proficiency with Python.
• Science: Deep expertise in at least one physical-science specialization.
• Research: Hands-on experience with computational modeling, simulation, numerical analysis, data analysis, or reproducible research.
• Evidence: Public technical work such as computational publications, preprints, public simulation or analysis code, research profiles, open-source contributions, reproducible notebooks, models, datasets, benchmarks, or experiments.