What the work actually is

You are not reviewing model output all day. You are authoring the problems. Each task is a self-contained piece of backend engineering — implement a feature against an existing codebase, or reproduce and fix a real bug — written so that a strong model finds it genuinely hard rather than trivially pattern-matched. The second half of every task is the harder half: a deterministic verifier. Usually a test suite, sometimes a harness plus fixtures, which must accept the full space of reasonable valid solutions and reject everything else. A verifier that only passes your own implementation is a failed task. So is one that passes a stub returning hardcoded values.

Expect to spend real time on edge cases, setup reproducibility, and pinning down what "correct" means when three different architectures all satisfy the spec. Languages in scope include Java, Node.js, Go, Rust, TypeScript, Python and C++. A cybersecurity or appsec background — secure code review, OWASP/CWE familiarity, vulnerability assessment, pentesting — is explicitly called out as a significant plus, since security-flavoured bug-fix tasks are in demand and hard to author well.

What the screen looks for

micro1 runs an AI interview of roughly 30 minutes, with a technical assessment that may follow, then hiring manager review. The interview probes depth: it will ask you to name systems you built, then follow up on the specifics — throughput numbers, what broke, how you diagnosed it. Vague seniority claims collapse under the second question. The platform asks that you complete both stages without outside assistance, including AI tools, and the same expectation carries onto the project itself. That request is not decorative; the entire value of the output is that it reflects your own judgment rather than a model's.

Logistics and pay

  • Contractor, fully remote, asynchronous. You choose hours and days, weekends included.
  • Roughly 15 hours a week, with a weekly minimum number of submitted tasks.
  • Compensation is output-based — paid per task that meets spec, not per hour. The $30–100/hr band is an observed effective rate and depends heavily on how fast you author and how often your tasks clear review on first pass.
  • Hiring moves quickly: roles often fill within 48 hours, and first tasks are expected within 24–48 hours of onboarding.