Human Resources Expert

Role Overview

We are building a network of Human Resources and People-operations experts to help train and evaluate frontier AI systems on real-world HR work. As a Human Resources Expert, you will apply your professional judgment to author, review, and stress-test tasks spanning the full people function — talent acquisition, compensation & benefits, HR business partnering, employee relations, learning & development, HR compliance, and HR analytics. Your expertise directly shapes how advanced AI models reason about the people decisions that organizations face every day.

Key Responsibilities

• Design realistic HR scenarios and tasks drawn from your hands-on experience across talent acquisition, comp & benefits, HRBP, employee relations, L&D, HR compliance, and people analytics.
• Craft model-facing prompts and author gold-standard responses that reflect sound, defensible HR practice.
• Review and grade AI-generated outputs for accuracy, completeness, and alignment with professional HR standards and applicable employment regulations.
• Identify subtle errors, edge cases, and failure modes in model reasoning on nuanced people decisions.
• Provide clear, structured written rationale to explain why a given answer is correct, incomplete, or wrong.
• Collaborate with the team to refine evaluation rubrics and raise the quality bar for HR domain content.

Ideal Qualifications

• 3+ years of professional experience in Human Resources or People operations (in-house, consulting, or agency).
• Depth in one or more HR sub-domains: talent acquisition/recruiting, compensation & benefits, HR business partnering, employee relations, learning & development, HR compliance, or HR analytics.
• Strong command of core HR frameworks, employment-law fundamentals, and people-decision tradeoffs.
• Excellent written communication — able to explain nuanced reasoning clearly and concisely.
• Sharp attention to detail and comfort evaluating the quality of others' work.
• An HR credential (SHRM-CP, SHRM-SCP, PHR, or SPHR) is preferred but not required.