The Work
You will support AI training focused on civil and commercial litigation strategy. The work uses litigation documents, case materials, and legal research to help AI systems learn from structured legal expertise and clear written feedback.
You will review source materials and model outputs against project guidelines, improve expert responses, explain your reasoning, and maintain consistent quality across repeated review tasks.
- Write or improve expert responses about civil and commercial litigation strategy.
- Review legal source materials and AI-generated outputs.
- Explain your judgments clearly so feedback can improve model behavior.
- Maintain accuracy and consistency across multiple review assignments.
What It Pays and Takes
This is flexible, part-time contract work. Time requirements vary based on experience and workflow. The role is open to candidates in the specified countries, including the United States, Canada, the United Kingdom, Australia, and countries across Europe, Asia, Africa, and the Middle East.
- $80 to $170 per hour equivalent, paid per task that meets the project specifications.
- Professional fluency in English is required.
- Juris Doctor degree from an accredited law school.
- Active law license in at least one U.S. jurisdiction.
- Extensive experience in civil or commercial litigation, including case strategy, discovery, and motion practice.
- Strong legal research, writing, analysis, communication, organization, and attention to detail.
- Experience consulting or training others on litigation topics is helpful.
How it works
You apply here in about two minutes and finish on the hiring platform. The next step is a short AI interview (technical roles include a coding challenge). Passing it is how you qualify for paid tasks, so take it soon after you are invited.
About AI Training Work
AI training is the human work behind systems that generate and evaluate language. Legal experts review examples, write strong responses, and explain decisions so AI models can produce more accurate and useful results.