The work

You will spend most of your time doing two things: authoring evaluation tasks that a capable AI model should be able to reason through, and grading model or draft outputs against your own expert determination. Source material is aggregate safety documentation — Development Safety Update Reports, PSURs/PBRERs, signal evaluation summaries, line listings, and the case-level records behind them. A typical task might ask you to check whether cumulative and interval exposure figures reconcile across sections and appendices, whether a benefit-risk conclusion is actually supported by the data presented, or whether an expectedness determination is consistent with the reference safety information.

Write-ups matter as much as verdicts. A finding of "case counts don't reconcile" is only useful to the customer if you specify which sections, which figures, the likely source of the discrepancy, and whether it would be material to a regulatory reviewer. Expect to produce structured rationales, not checkbox scores. Reviewers on the project will push back on reasoning that is correct but unexplained.

What the screening measures

micro1's process is AI-led, with a conversational interview followed by domain tasks. It is looking for evidence you have owned aggregate reports end to end rather than contributed to one section — so be ready to describe specific reports you authored or reviewed, the therapeutic areas, the regulatory context, and decisions you made under time pressure. Expect follow-up questions that go a layer deeper than your first answer, particularly on ICH E2F versus E2C(R2) structural requirements, signal management workflow, MedDRA coding judgment, and how you handled a discrepancy you could not fully resolve. Writing clarity is assessed directly; short, vague answers screen poorly regardless of credentials.

Logistics

  • Contractor engagement, fully remote, no fixed hours.
  • Asynchronous task queues with per-batch deadlines; most contributors work in blocks of a few hours rather than full days.
  • Observed pay band for this listing is $70–80/hr, stated as observed and not guaranteed — actual rate depends on assessed depth, task type, and project stage.
  • Volume fluctuates with customer demand; treat it as supplemental rather than a replacement for primary work.
  • No prior AI or annotation experience expected.