AI Trainer (Physical Sciences)

Micro1; India; Remote; INR 7,768,000 - 15,537,000; Part time

Job summary

Cashback24x7 seeks Computational Physical Scientists to support a project focused on advancing AI systems through realistic physical-science tasks. The role is fully remote and global, offered as a contractor with flexible schedule and a 15+ hour weekly commitment.

You will create tasks, reference environments, evaluation criteria, and solutions, then evaluate model results, identifying errors and reproducibility issues.

Qualifications

• Paid experience training, evaluating, benchmarking, or improving AI systems on scientific or technical tasks.
• Strong practical proficiency with Python.
• Deep expertise in at least one relevant physical-science specialization.
• Hands-on experience using computational methods for modeling, simulation, numerical analysis, data analysis, or reproducible research.

Responsibilities

• Create realistic computational physical-science tasks that challenge advanced AI models.
• Develop task instructions, reproducible environments, reference solutions, and evaluation criteria.
• Evaluate model-generated calculations, code, simulations, models, and technical deliverables.
• Identify invalid assumptions, numerical errors, physically impossible results, incorrect methods, or reproducibility failures.

Skills

Python

Physics

Computational methods

Data analysis

Reproducible research

NumPy

SciPy

SymPy

Astropy

PyTorch

JAX

Tools

NumPy

SciPy

SymPy

Astropy

PyTorch

JAX

Job description

Pay

Pay: $60-$120/hour equivalent, paid per accepted task

Location

Location: Global, fully remote

Job Type

Job Type: Contractor

Time Commitment

Time Commitment: 15+ hours per week

Schedule

Schedule: Flexible

Project Overview

micro1 is looking for Computational Physical Scientists with prior professional AI-training or model-evaluation experience to support a project focused on improving advanced AI systems.

Responsibilities

• Create realistic computational physical-science tasks that challenge advanced AI models.
• Develop task instructions, reproducible environments, reference solutions, and evaluation criteria.
• Evaluate model-generated calculations, code, simulations, models, and technical deliverables.
• Identify invalid assumptions, numerical errors, physically impossible results, incorrect methods, or reproducibility failures.

Required Qualifications

• Prior paid experience training, evaluating, benchmarking, or improving AI systems on scientific or technical tasks.
• Strong practical proficiency with Python.
• Deep expertise in at least one relevant physical-science specialization.
• Hands-on experience using computational methods for modeling, simulation, numerical analysis, data analysis, or reproducible research.
• Relevant specializations may include physics, astronomy and cosmology, materials science, chemistry, computational physics, computational chemistry, and scientific simulation.
• Public Proof of Skill: Strong applicants should provide evidence such as: Peer-reviewed publications or preprints involving computational methods, Public simulation or analysis code, Google Scholar, ORCID, arXiv, or equivalent research profiles, Open-source scientific software contributions, Reproducible notebooks, models, datasets, or simulations, Public benchmarks, experiments, or technical projects.

Preferred Qualifications

• Experience creating AI-training tasks, reference answers, rubrics, graders, or benchmark problems.
• Experience with NumPy, SciPy, SymPy, Astropy, PyTorch, JAX, or comparable scientific tools.
• Experience validating simulations or models against analytical results, observations, or experimental data.
• Availability to begin immediately.

Compensation

Compensation is output-based. Experts are paid for tasks that meet project specifications and quality requirements. Task availability and completion time may vary.