Materials Characterisation Scientist, AI Evaluation

You will create research-level problems that test and train AI models. Your work will use real instrument data and your knowledge of materials science, metallurgy, and laboratory characterisation.

You will need to analyze technical data, explain your reasoning clearly, and document accurate, technically sound content.

• Write research-level problems based on materials science or metallurgy expertise.
• Develop problems using real XRD, indentation, impedance, corrosion, rheology, or battery cycling data.
• Analyze instrument data with Python or a similar programming language.
• Keep technical content accurate, clear, and well documented.

What it pays and takes

• Pay: $50 per hour.
• Work type: Flexible, task-based independent-contractor work with no fixed weekly schedule.
• Location: Remote and open to candidates in the countries listed for this role.
• Language: Fluent English.
• Required: Hands-on experience working with your own materials characterisation data.
• Required: Working knowledge of Python or a similar language for instrument-data analysis.
• Required: Experience with laboratory characterisation in materials science, metallurgy, or a related field.
• Required: Clear technical writing, independent working skills, and the ability to manage your own time.
• Helpful: A Master's or PhD in materials science, metallurgy, or a related field.
• Helpful: Experience with Rietveld, Oliver-Pharr, EIS fitting, ASTM test standards, analysis scripting, or technical work shown through a thesis, paper, or public code.

How it works

Apply on OpenTrain with your resume, then complete the application on the hiring site. OpenTrain is where you find this role and start the application; it does not employ or pay contributors for the work.

About AI training work

AI training work uses examples and expert feedback from people to improve how artificial intelligence systems respond. Specialists are paid for applying subject knowledge, reviewing technical material, and creating reliable examples that help models perform better.