Litigation and Civil Procedure AI Evaluation Expert

You will evaluate AI-generated answers to litigation hypotheticals, using your knowledge of civil procedure, evidence, and trial strategy. Your assessments and written feedback will help improve language models.

• Review responses about motions to dismiss, summary judgment, discovery disputes, and trial strategy.
• Use a structured legal rubric to assess issue spotting and accuracy under the Federal Rules of Civil Procedure, case law, and evidence rules.
• Identify gaps involving standing, jurisdiction, burden of proof, Twombly and Iqbal pleading standards, and remedies.
• Write precise annotations and feedback, and work with researchers and project managers to align evaluation standards and project goals.

What it pays and takes

• Pay: Not provided.
• Hours: 20 or more hours per week.
• Location: Fully remote and open worldwide.
• Engagement: Freelance work expected to last about one month, with a possible extension based on performance and project needs.
• Education and admission: Juris Doctor degree from an accredited U.S. law school and active or inactive bar admission.
• Legal experience: At least three years in U.S. civil litigation, commercial litigation, or appellate work.
• Skills: Strong knowledge of the Federal Rules of Civil Procedure, evidence, and trial advocacy; clear writing, analytical reasoning, attention to detail, and ethical judgment.
• Helpful background: Experience analyzing litigation strategy, identifying procedural issues, and explaining legal conclusions clearly. Prior AI experience is not required.
• Language: Fluent in English.

How it works

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

About AI training work

OpenTrain is a platform for finding AI-training and data-labeling roles and building a profile to support your applications. In AI training, people evaluate and annotate examples so AI models can learn from human feedback; legal expertise is valuable for judging complex legal responses.