What the work involves

micro1 recruits AI engineers for two overlapping streams of work, and which one you land in depends on how the screen goes. The first is conventional contract engineering: building and shipping LLM-backed features — retrieval pipelines, agent loops, tool-calling layers, evaluation harnesses, fine-tuning and serving infrastructure — usually embedded with a client team on a multi-month engagement. The second is expert data and evaluation work for AI labs: writing reference solutions to hard engineering problems, constructing test cases that break models, grading model-generated code against a rubric, and writing the critique that explains why an output fails rather than just scoring it.

The listing as published carries no scoped description, which is normal for micro1 — matching happens after the screen rather than before it. Expect the recruiter conversation to sort you toward a specific client or lab project based on what your interview surfaces. Pay in the $60–120/hr range has been observed across these engagements; the top of the band tends to attach to full-time contract placements with production ownership, the lower end to part-time evaluation and data work.

What the platform screens for

  • Real code under observation. The AI interviewer runs a live coding segment. It watches your process, not just the final answer, and will interrupt with follow-ups.
  • Depth on claims. Anything on your resume is fair game for three or four layers of drill-down. Vague ownership claims collapse quickly under this format.
  • Production judgment with LLMs. Not prompt trivia — questions about latency budgets, cost, eval design, failure modes, and when not to use a model.
  • Written reasoning. For eval-track work, the ability to articulate why an output is wrong in specific, actionable terms.

Logistics

Fully remote and global; micro1 hires across time zones. Contract placements typically expect substantial overlap with a US or European client team — often four hours or more — and 20–40 hours a week. Evaluation and data projects are more async and can run at 10–15 hours weekly. Onboarding includes identity verification and, for client placements, a background check. Work is invoiced through the platform.