What the work involves

You are handed code patches an LLM produced for a repository you know intimately — ideally one you maintain or contribute to regularly. Your job is to judge each patch the way you would judge a drive-by pull request from an unfamiliar contributor: does it compile, does it pass the suite, does it solve the stated issue, does it match the project's conventions, and does it introduce a subtle regression that tests do not catch? Each verdict ships with written justification. Vague ratings are the single most common reason submissions get rejected; the value being purchased here is your reasoning, not your score.

Beyond first-pass review, you will peer-review assessments submitted by other subject-matter experts. That means disagreeing on the record when someone has waved through a patch that quietly changes error semantics, or flagged a stylistic choice as a bug.

What the platform screens for

  • A real, inspectable open source footprint. micro1 states plainly that applicants need clear open source contributions with a GitHub or GitLab profile to show it. Commit history, merged PRs, issue triage, and review comments on other people's code all count; a profile of forks and tutorial repos does not.
  • Maintainer instincts over language trivia. Preference goes to people who have owned a repo — set contribution guidelines, cut releases, declined patches, handled a CVE report.
  • Written review quality. Expect the AI interview to push you to justify a judgment and then push again on the edge case you skipped.

Logistics

Contractor engagement, roughly 15 hours per week, remote and open globally with no fixed hours — the stated expectation is constant activity rather than a schedule. Compensation is output-based: you are paid per task that meets project specifications, with a weekly minimum submission requirement, so the observed hourly band reflects throughput and varies with your workflow. The pipeline is screening questions, a roughly ten-minute AI interview, then hiring manager review; roles typically fill within 48 hours and selected experts are expected to start their first tasks within 24–48 hours of onboarding.