The work
You will evaluate AI-generated mathematics answers and decide whether their calculations, reasoning, and explanations are correct. You will also write or refine mathematical explanations and model solutions.
The work covers pure and applied mathematics, proofs, modeling, probability, statistics, and optimization.
- Review AI-generated math answers for correctness, reasoning quality, and clarity
- Identify calculation errors, gaps in methodology, and conceptual mistakes
- Check calculations, proof structure, methods, and conceptual validity
- Fact-check quantitative claims
- Write and refine mathematical explanations and model solutions
- Compare and rank AI responses based on mathematical correctness and reasoning quality
What it pays and takes
This is a part-time contractor role listed as entry level. Prior data-labeling, reinforcement learning from human feedback, or AI model evaluation experience is helpful but not required.
- Pay: $70 per hour
- Time: 20+ hours per week
- Location: Open to candidates in Bangladesh, Bhutan, Brazil, Cambodia, Germany, India, Indonesia, Malaysia, Nepal, Pakistan, Singapore, Sri Lanka, Thailand, the Philippines, the United States, Timor-Leste, and Vietnam
- Language: Fluent English with excellent writing and clear mathematical communication
- Education: MS or PhD in mathematics, statistics, or a related field from a top 100 university
- Knowledge: Strong command of pure and applied mathematics, proofs, modeling, probability, statistics, and optimization
- Background: Research, analytical writing, debate, programming, or mathematics experience
- Helpful experience: Developing or critically reviewing problem banks, proofs, textbook sections, research notes, or other complex math content
How it works
Apply on OpenTrain with your resume, then complete the application on the hiring site.
About AI training work
AI training is the human work behind systems that learn from examples, including reviewing model answers and rating their quality. People with strong subject knowledge help identify errors and create better examples for AI systems to learn from.