What the work actually looks like

A task typically arrives as a self-contained problem: a terminal environment, an unfamiliar repository or a precise specification, and a goal stated tightly enough that correctness can be checked automatically. You read the objective, find the ambiguity in it, choose an algorithm and data structure that fit the stated constraints, implement it, and then test against conditions that may be hidden from you. Some tasks are pure algorithmic reasoning; others are debugging — behavior that is wrong because of the code, the tests, a dependency, or the environment configuration, and your job is to isolate which. A significant share involves reviewing AI-generated solutions and identifying where the logic breaks, where the reasoning stops short, or where a constraint was quietly violated by code that looks entirely reasonable.

The deliverable is never just a passing answer. It is a solution someone else can execute, plus tests that separate a genuinely correct implementation from a plausible-looking one, plus a written explanation of why you made the trade-offs you made. That last part carries real weight: these artifacts become training and evaluation signal, and an unexplained correct answer is worth less than a well-reasoned one.

What the platform screens for

micro1 runs an AI-led screen of roughly 30 minutes, potentially followed by a technical assessment and a hiring manager review. Expect the screen to push on depth rather than breadth — complexity reasoning under follow-up, how you narrow a bug in code you have never seen, how you decide a requirement is ambiguous rather than guessing. Competitive-programming history helps and a verifiable Codeforces profile is explicitly listed as nice to have, but it is not a gate; demonstrable fluency in one general-purpose language and clear reasoning about correctness matter more.

Logistics

  • Fully remote, global, contractor engagement at roughly 15 hours per week.
  • Fully async and self-scheduled, weekends included; no fixed meetings stated.
  • Compensation is output-based — paid per task meeting spec, with minimum submission requirements. The $50–80/hour band is the observed effective rate and varies with how quickly you work.
  • Roles are typically filled within about 48 hours, with the first task expected 24–48 hours after onboarding.