This companion looks at one practical step before you apply or accept work under any of the six micro1 listings that this site groups as generalist-evaluator work. It does not repeat the overview in the Main generalist-evaluator guide; it isolates one decision point: figuring out, from each listing’s own scope language, whether the work is mostly authoring new material, mostly reviewing material already produced, or a mix of both — and what to ask in writing before you accept a task paid under an output-based structure.
The full descriptions behind this page were read directly on jobs.micro1.ai on 2026-10-01 and saved to this site’s research file. That date does not confirm that any of the six listings is open today, what it currently pays, who is eligible, or what today’s interview steps are. Treat everything below as a record of what those pages said on that date, not as a live posting.
Authoring or reviewing: read the scope line first
Before drafting a profile or submitting a screening form, read the “Scope of Work” section of the listing word for word and sort it into one of three buckets:
- Mostly authoring. The Research Evaluation Specialist listing and the Evaluation Specialist/Recent Grad listing both describe writing original question-and-answer pairs, sourcing and documenting the answers, then testing those questions against AI systems and adjusting their own difficulty — that testing step is aimed at the author’s own material, not at reviewing output someone else produced.
- Mixed. The STEM Expert listing and the Technical Problem Author listing both describe two separate steps done by the same person: authoring a problem (a scientific problem with a reference solution, or a technical problem in an engineering domain), and then, as a distinct task, reviewing an AI-generated response to it for correctness.
- Mostly reviewing. The AI Domain Expert listing describes reviewing and critiquing AI-generated outputs, designing prompts, and giving written feedback, without describing authorship of new problems from scratch. The Gmail & Google Calendar AI Assistant Evaluator listing describes completing an everyday task through an AI assistant and then applying grading guidelines to rank and annotate what the assistant produced.
This distinction matters for preparation: a listing built around authoring rewards practice researching and sourcing a claim; a listing built around reviewing rewards practice reading someone else’s work closely against stated criteria; a mixed listing asks for both skills in the same role. Confirm which bucket a specific listing falls into before deciding how to prepare — the title alone (“Evaluator,” “Expert,” “Author”) does not reliably tell you.
Payment: what these six listings do and do not say
None of the actual figures are repeated here; the point is what each listing states or omits, not the numbers themselves.
- Advertised range. Only the STEM Expert listing’s own page states an hourly figure. The other five saved descriptions do not state a pay range.
- Paid unit. The STEM Expert, Research Evaluation Specialist, Evaluation Specialist/Recent Grad, and Technical Problem Author listings all state that compensation is output-based and paid per task that meets project specifications. The AI Domain Expert and Gmail & Google Calendar AI Assistant Evaluator listings do not state a paid unit in the saved text.
- Minimums. The same four output-based listings state that a minimum submission requirement applies, without the saved text giving the number. The other two listings do not mention a minimum.
- Availability. The STEM Expert listing states a weekly-hour figure for the role; the saved text for the other five listings does not state one.
- Acceptance criteria. The four output-based listings say a task must “meet project specifications” without defining that phrase. The Gmail & Google Calendar listing says evaluators apply grading guidelines to rank and annotate an assistant’s work, but the saved text does not say who makes the final accept-or-reject call. The AI Domain Expert listing does not describe an acceptance step at all.
None of the six saved descriptions explain what happens to a submission that is sent back for revision, or whether a reviewed-but-rejected task still counts toward a weekly minimum. Before accepting work, put these in writing:
- What exactly counts as the paid unit for this specific task — a single question-and-answer pair, a single reviewed response, a single completed scenario?
- Who accepts the work, and by what stated criterion?
- What happens to work that is returned or rejected — is it resubmitted, replaced, or simply unpaid?
- What is the minimum, in concrete terms, for this role?
Ask micro1 for the payment conditions in writing before accepting work. Because pay in the four output-based listings is tied to the task, not the clock, extra time spent on revisions or clarifications changes your effective pay per hour of work, even though it does not change the stated price of the task itself. The saved descriptions do not say how retries are counted, which is exactly why this belongs on your written-questions list rather than in your assumptions.
Practice drafting the clarifying message
The examples below are illustrative exercises created for this guide, not published micro1 assignments or interview questions.
Each draft below is a fictional, ungraded message a reader might send after reading a specific listing’s scope section — not a required script, and not an example of company wording.
Draft 1 — addressed to a Research Evaluation Specialist-type listing: “Before I submit a sample item, could you clarify what counts as ‘meeting the project specifications’ for a single question-and-answer pair, and whether a pair that needs a revision round still counts toward the weekly submission minimum once it’s approved?”
Draft 2 — addressed to a Technical Problem Author-type listing: “The listing separates authoring a problem from reviewing an AI-generated response to it. Are these treated as two separate paid tasks, or as one combined submission?”
Draft 3 — addressed to a Gmail & Google Calendar Evaluator-type listing: “If a grading pass flags the assistant’s output as needing a redo, does that redo count as part of the same submission, or as an additional one?”
Checklist before you apply or accept
- Read the scope section of the specific listing and sort it into authoring, reviewing, or mixed — don’t rely on the job title.
- Note whether the listing states a paid unit, a minimum, or an acceptance step; mark anything it does not state as unknown, not as absent.
- Draft your own written questions before applying, covering the paid unit, who accepts the work, what happens to returned work, and the minimum.
- Keep any written answer micro1 gives you about these points; the saved listing text does not substitute for it.
- Revisit this checklist for each listing separately — the output-based listings share language, but the AI Domain Expert and assistant-evaluator listings do not describe the same payment structure.
For the fuller picture of what each of these six roles asks for and what it does not say, see the Main generalist-evaluator guide.