What the work actually involves

You will spend most of your time producing written legal reasoning that a model can learn from. In practice that means taking a custody, support, property division, or premarital agreement fact pattern and writing out the analysis the way you would for a partner or a client memo — statute, controlling authority, the argument, the counterargument, and where the answer turns on jurisdiction or judicial discretion. Other tasks include reviewing model-drafted separation agreements or parenting plans and marking exactly where the drafting fails, comparing two model answers on the same prompt and defending which is better, and writing short case notes that isolate the operative issue from the noise.

The recurring difficulty is that family law is state-specific and equitable. A model that confidently states a single national rule on relocation, imputed income, or the enforceability of a prenup is producing exactly the failure mode the project exists to correct. Reviewers who flag jurisdictional overreach and unstated discretionary standards are more valuable here than reviewers who only catch citation errors.

What the screen looks for

micro1 runs an AI-led interview. It checks that your BigLaw family law experience is real and specific — which firm, which matters, which states you practiced in, what you personally drafted versus supervised — and then probes depth with follow-ups. Expect to be pushed on a doctrinal answer you give, and expect the follow-up to test whether you can say "that depends on the forum, and here is why." Written clarity matters as much as doctrine; the deliverable is prose, not a conclusion.

Logistics

  • Fully remote, contractor engagement, no AI or ML background required.
  • Largely asynchronous work through a project platform, with deliverables on defined turnarounds.
  • Volume varies by project phase; many contributors work part-time alongside practice, commonly in the 10–20 hour range per week.
  • Observed pay band is $140–400/hr, set by seniority and task type. Rates are as observed on the platform, not guaranteed.