What the work involves
Day to day, you work through queues of prompts and model responses that mix Spanish-language instructions with code. Typical tasks include writing prompts a real developer would ask in Spanish, judging whether a model's answer is correct and runnable, rewriting explanations so the technical vocabulary is right in Spanish rather than a literal translation of English, and comparing two candidate answers against a rubric with a written justification for the choice. Some projects add error-labelling — tagging whether a failure is a factual mistake, a broken code block, a translation artefact, or a refusal that should not have happened.
The rubric matters more than personal taste. Reviewers are checking that your labels are reproducible: two competent annotators reading the same guideline should land on the same answer, and where they wouldn't, you are expected to flag the ambiguity instead of quietly picking one. Guidelines get revised mid-project, and re-reading a changed spec carefully is part of the job.
What the screen looks for
micro1 runs an AI-led interview before any human conversation. Expect it to probe two things at once: genuine Spanish fluency at a professional register, and real programming experience it can push on with follow-up questions. Vague claims get followed up rather than accepted, so name languages, projects, and the specifics of what you built. You may be asked to reason aloud about a flawed answer or to explain what you would do when a rubric doesn't cover the case in front of you.
Logistics
- Remote, largely asynchronous, with queues you clear against weekly targets
- Monthly rates in the $3,000–5,000 range have been observed on comparable micro1 engagements; the actual figure depends on hours committed, seniority and project, and is not guaranteed
- Contractor engagement, usually with an expectation of steady full-time or near-full-time availability and some overlap with the project team's calls
- Reliable internet and a personal machine are assumed