What the work involves

You write and review code the way a senior engineer would on a real project, but the artifact is training and evaluation data rather than shipped software. A typical task might ask you to implement a service in Go against an ambiguous spec, then write the reasoning trace explaining your design choices; or to take a model-generated Python module and produce a detailed critique covering correctness, idiomatic style, error handling, and the edge cases the model missed. Other tasks lean toward unit tests, documentation, or writing prompts that will reliably separate strong model outputs from plausible-looking wrong ones.

  • Author idiomatic Go, with Python and TypeScript as secondary languages
  • Write structured written feedback on sample code — specific, cited to lines, not general impressions
  • Produce tests, docs, and edge-case inventories for design and concurrency problems
  • Refine prompts and rubrics used to grade AI-generated code

What the platform screens for

micro1's screen is AI-led and conversational, followed by technical exercises. Expect it to probe your actual Go experience under follow-up: goroutine lifecycle and cancellation, interface design, error wrapping, module and build tooling, and the tradeoffs you made on specific systems you shipped. Secondary-language questions test whether you write clean, typed Python and TypeScript rather than whether you can name frameworks. Because the deliverable is largely prose, written clarity is weighted heavily — vague critiques and unfalsifiable claims are the most common rejection reason. No prior AI or ML background is required or expected.

Logistics

Fully remote, contractor engagement, asynchronous with no fixed shifts. Most contributors commit 10–20 hours per week and are asked to be realistic about sustained throughput rather than optimistic. Pay in the $100–130/hr range has been observed on this listing and typically tracks language depth and calibration on review tasks; rates are set per project and are not guaranteed. Expect a short calibration period where early submissions receive reviewer feedback before volume opens up.