Political Science Domain Expert

You will help assess and improve large language models using your knowledge of political science and related fields. The work combines research judgment with prompt creation, response evaluation, benchmark development, and evidence-backed feedback.

• Write advanced prompts about political science, governance, elections, public policy, and international relations.
• Rate AI responses for factual accuracy, reasoning, completeness, nuance, and current information.
• Find hallucinations, logical inconsistencies, outdated claims, and difficult edge cases.
• Create benchmark datasets and adversarial test cases for challenging model evaluations.
• Give objective feedback supported by reliable references.
• Work with AI researchers to improve model performance and maintain clear annotation documentation.

This is a contractor assignment scheduled for eight weeks. The role details do not list a pay rate.

• Time requirement: 40 hours per week, including at least four hours of overlap with Pacific Time.
• Location: Open worldwide.
• Language: Excellent written English is required.
• Experience level: Entry level.
• Education: A master's degree or higher in any field is required.
• Relevant experience: At least three years in political research, public policy, government, political consulting, academia, journalism, international affairs, think tanks, or a related area.
• Knowledge: Political institutions, comparative politics, governance, elections, public policy, and current global political developments.
• Skills: Research, analysis, documentation, careful review, and the ability to identify factual inconsistencies and assess reasoning across political topics.
• Helpful background: Experience with large language models, generative AI, prompt engineering, or AI evaluation.
• Additional strengths: Published research, teaching experience, or industry recognition.

Apply on OpenTrain with your resume and then complete the application on the hiring site.

AI training is the human work behind systems that learn from examples, including writing and rating model responses. People with strong subject knowledge help make evaluations accurate, fair, well-supported, and useful for improving AI systems.