Domain Expert - History

Role Overview

We are seeking a highly qualified History Domain Reviewer to support the quality assurance of Large Language Model (LLM) evaluation projects. In this role, you will review domain-specific prompts, evaluate completed tasks for accuracy and quality, ensure adherence to project guidelines, and provide actionable feedback to maintain high annotation standards. Your expertise in history will help ensure the reliability, consistency, and quality of AI evaluation data across diverse historical topics.

Key Responsibilities

• Review and validate domain-specific prompts covering ancient, medieval, modern, and contemporary history, civilizations, historical events, political history, military history, archaeology, historiography, and related topics.
• Evaluate completed tasks to ensure factual accuracy, reasoning quality, completeness, and compliance with project guidelines.
• Identify factual inaccuracies, logical inconsistencies, hallucinations, outdated information, and low-quality annotations.
• Ensure prompts are challenging, relevant, and aligned with project objectives.
• Provide clear, constructive, and evidence-based feedback to contributors to improve task quality.
• Maintain consistency across reviews by following established quality standards and review guidelines.
• Escalate ambiguous or complex cases when necessary and document review findings.
• Collaborate with project managers and AI teams to continuously improve evaluation quality and review processes.

Minimum Qualifications

• Master's degree or higher in History, Archaeology, Historical Studies, Political History, World History, or a closely related discipline.
• Minimum 3 years of professional, research, teaching, or industry experience in the history domain.
• Prior experience reviewing, evaluating, editing, or quality-checking domain-specific content is preferred.
• Excellent written English communication and analytical skills.
• Strong attention to detail and the ability to identify factual, logical, and contextual errors.