Science Reasoning Expert — STEM Problem Author

About OpenTrain

OpenTrain is the centralized platform where people build careers in AI training and data labeling. We help contributors find projects, consolidate opportunities, and grow a unified portfolio of AI-training work they control.

OpenTrain AI is hiring directly for this contract role. Our work connects domain experts with high-impact tasks that shape how state-of-the-art AI systems reason and solve problems.

About AI training and why it matters

AI training (also called data labeling or human feedback work) is the human side of building intelligent systems: people create, review, and rate examples that teach models to reason, write, and solve problems. This field offers flexible, remote work that directly influences how models behave.

Experts who author and evaluate complex scientific problems help set the standard for correctness and rigor in model reasoning—the work is highly impactful for research-quality systems and applied AI alike.

The role — Science Reasoning Expert

You will design and review challenging, cross-disciplinary STEM problems and evaluate AI-generated solutions for correctness, rigor, and clarity. Tasks are text-based: authoring questions, reviewing peer content, and rating model outputs with written justification.

This is a contract, part-time role requiring 20+ hours per week. Work is remote and worldwide; English fluency is required. OpenTrain AI will pay $75 USD per hour.

• Job type: Contractor, Part-time
• Minimum weekly time: 20+ hours
• Pay: $75 USD per hour
• Data type: Text (problem writing and evaluation)
• Labeling tasks: Text generation and evaluation/rating

What you'll do day to day

You will both create and critically evaluate high-difficulty STEM problems and model solutions. Expect deep, written feedback and clear pass/revise/reject decisions grounded in scientific reasoning.

• Design expert-level problems across physics, chemistry, biology, and mathematics, including olympiad-style and discovery-focused tasks.
• Review peer-authored problems for clarity, genuine difficulty, and correctness before they enter datasets.
• Evaluate AI-generated solutions and multi-step reasoning traces; deliver Accept, Revise, or Reject with detailed written rationale.
• Document correct solutions and edge-case notes that define ground-truth answers for training data.

Requirements

You must have an expert-level, broad STEM background and proven experience reasoning through multi-step scientific problems. Strong scientific writing and attention to detail are essential because evaluations require written justification of correctness.

• Broad expertise across physics, chemistry, biology, and mathematics (required).
• Ability to judge multi-step scientific reasoning outside a single specialty.
• Experience authoring or reviewing difficult STEM problems.
• Strong scientific writing and meticulous attention to detail.
• Science olympiad medals, competition wins, or similar high-level research credentials are strongly preferred.
• Fluent English (reading and written) and availability for 20+ hours/week.

Who should apply

This role suits researchers, competition-level problem solvers, and senior students or educators with cross-disciplinary STEM strength who enjoy crafting and critiquing hard problems and evaluating reasoning traces.

If you value rigorous standards, clear written feedback, and flexible remote contract work, this role offers direct influence on how AI models learn scientific problem solving.

Compensation, logistics, and how to apply

Compensation is $75 USD per hour on a contractor basis. The role is remote and open worldwide. Assignments are text-based and managed through OpenTrain’s systems.

To apply, create or update your OpenTrain profile, list relevant STEM credentials and sample problems or evaluation notes, and submit an application. Short expert screening tasks may be used to confirm fit and judgment quality.

• Rate: $75/hr (USD) — contractor pay
• Hours: 20+ hours/week, part-time scheduling and remote work
• Process: profile, application, expert screening tasks to demonstrate problem-design and evaluation skills