You will turn legal work you know well into realistic simulations used to train and evaluate AI systems. Each task packet includes a practical request, a file set, an answer key, and a scoring rubric.
The people, matters, and documents in each simulation are fictional. You will write from your own experience while adding realistic details such as superseded drafts, unhelpful emails, and missing documents.
- Write the request as it would arrive from a partner, general counsel, or client.
- Build the day-one file set, including agreements, correspondence, pleadings, discovery, matter files, policies, or prior work product.
- Write the answer key and explain the reasoning behind it.
- Design a scored rubric that separates required points, serious errors, and defensible judgment calls.
- Grade answers from other authors or AI systems without seeing the author and identify problems with the rubric itself.
- Join calibration sessions where authors grade the same answer and reconcile their decisions.
- Map one familiar workflow, including its steps, inputs, decision points, five common reasons work is sent back, and sources a new hire should consult.
Legal areas and task format
Your area of law is open. Depth in one practice area matters more than broad coverage. A typical packet takes 2 to 4 hours to create.
- Litigation support
- Corporate and transactional work
- Commercial contracts
- Privacy and data protection
- Employment
- Regulatory and compliance work
- Intellectual property
- Real estate
- Trusts and estates
- Tasks may involve legal analysis, drafting, question answering, and evaluation or grading.
What it pays and takes
This is a remote, part-time contractor role for experienced legal practitioners who can explain why work is correct or incorrect in writing. You must be comfortable using a template and receiving detailed, line-by-line feedback from researchers who may not be lawyers.
- Pay: $200 per label.
- Time: 20 or more hours per week.
- Location: Open to candidates in the United States.
- Language: Professional fluency in English.
- Experience: Expert-level legal work experience.
- You must have at least 5 years of hands-on legal work as an attorney, paralegal, contracts manager, privacy or compliance analyst, or litigation-support specialist.
- You must have drafted, assembled, filed, redlined, or analyzed legal documents yourself, rather than only reviewing work for someone senior.
- You must have created at least one teaching or assessment tool, such as a training exercise, onboarding checklist, quality review form, playbook, continuing legal education material, law-school hypothetical, bar-prep question, or model answer.
- You must be able to explain legal errors precisely in writing so a junior can understand the issue and a partner can accept the explanation.
Helpful background
Experience supervising or training juniors, paralegals, or offshore teams is helpful. So is writing internal standards, review checklists, sample work product, hypotheticals, moot problems, exam questions, practical assessments, or AI-training tasks.
Application exercise
The application includes a short writing exercise that takes about 60 minutes and can be completed on your own time. You will receive a one-paragraph matter and prepare a file list, the request, and a ten-line rubric, followed by a 20-minute call.
Your application must also answer three questions about legal work you have performed repeatedly, something you created to teach or assess it, and a rubric line that would catch one common junior error.
- Name one piece of legal work you have done more than 50 times and its three most common junior errors.
- Describe a checklist, exercise, sample answer, or review form you created, who used it, and what changed.
- Write two or three sentences describing the rubric line that would catch one of those errors, including what the grader looks for and what evidence in the file supports it.
How it works
Apply on OpenTrain. The employer reviews applications there.
About AI training work
AI training work uses examples prepared and reviewed by people to help AI systems produce better results. Legal specialists are needed because strong evaluations depend on practical judgment, accurate reasoning, and clear standards for quality.