About the Role
Mercor is partnering with a leading AI lab to train frontier models on high-quality legal reasoning data. This is an evergreen talent pool for experienced US Litigation attorneys — by applying, you join a standing bench and are considered for ongoing and upcoming legal-AI projects as they launch. We review applications on a rolling basis and reach out when work that fits your background comes up.
Typical work includes recurring side-by-side evaluations and legal dataset generation: you'll compare AI-generated legal work, judge which response is stronger and why, and create realistic litigation tasks and reference answers.
What You'll Do
- Compare pairs of responses to litigation tasks (research memos, pleadings, motions, briefs, discovery and case strategy) and decide which is stronger, with clear reasoning
- Write concise justifications grounded in the source documents and applicable US law
- Create realistic legal prompts, fact patterns and model answers drawn from your litigation experience
- Flag errors, gaps and risky advice in AI-generated legal content
- Attend onboarding and calibration sessions
You're a Good Fit If You
- Are admitted to practice in at least one US state (active bar membership)
- Have 3+ years of US litigation practice, such as civil or commercial litigation, motion practice, discovery and trial preparation
- Hold a J.D. from an ABA-accredited law school
- Write clearly and pay close attention to detail