What the work actually involves

You'll spend most of your time on two intertwined activities: producing coding problems and solutions that are hard enough to differentiate strong models from weak ones, and judging model output against those problems. That means analysing and debugging unfamiliar codebases in Python3, Java, Rust, C++, Go or TypeScript; implementing features or refactors that serve as ground truth; and authoring test cases that catch the specific ways a model fails — off-by-one edges, misread constraints, plausible-looking solutions that TLE on the intended input size. Every task ships with written justification: why the reference solution is correct, why a model's attempt is wrong, and where exactly it went wrong. The annotation is the product as much as the code is.

What the screen is looking for

micro1's screening is AI-led and leans on verifiable specifics. Expect probes into competitive programming history (handles, ratings, platforms, problem-setting or testing credits), open-source contributions you can point to by repository and PR, and follow-up questions that test whether your algorithmic knowledge survives contact with a concrete constraint — complexity reasoning, edge-case enumeration, why one approach beats another at a given input bound. The listing states no prior AI experience is required; what it screens for is depth in the code domain and the ability to write about it clearly. Vague seniority claims do poorly here. Named problems, named repos, and stated reasoning do well.

Logistics and pay

  • Remote, asynchronous, contractor. No fixed shifts; you pick up tasks from a queue.
  • Output-based pay. You're paid per task that meets spec, not per hour logged. The $50–100/hr band is what contributors at this tier have observed once throughput stabilises — slower ramp-up periods effectively pay less, and rejected work pays nothing.
  • Weekly minimum. A minimum number of submissions per week applies; the exact figure is set per project.
  • Fast start. Roles are typically filled within 48 hours, with first tasks expected 24–48 hours after onboarding. If you can't begin almost immediately, say so in the screen rather than after selection.