What the work actually looks like

You are not being hired to build systems. You are being hired to demonstrate, in writing, how a competent systems analyst reasons — and to judge whether a model's attempt at the same task holds up. A typical block of work might be: read an ambiguous stakeholder request, write the clarifying questions you would actually ask, produce the requirements artifact that follows, and then critique two model-generated versions of the same artifact against yours. Other tasks are SQL-centric: given a schema and a business question, write the query, explain the join logic and the edge cases (nulls, duplicate grain, late-arriving rows), and mark exactly where a model's query silently returns the wrong number.

Process modeling shows up constantly. Expect to author BPMN diagrams as structured text or notation, review whether a model has correctly represented gateways, message flows, and exception paths, and explain why a swimlane assignment is wrong rather than just flagging it. The explanation is the deliverable — a rejection with no diagnosis is worth very little to the customer.

What the screen is looking for

micro1's screening is AI-conducted and follow-up heavy. It probes whether your five years are real: it will take a general claim about requirements elicitation and ask what you did when two stakeholders gave contradictory acceptance criteria on the same field. It tests SQL by discussion rather than by IDE — you should be able to talk through a query's grain, describe why a LEFT JOIN plus a WHERE clause on the right table quietly becomes an INNER JOIN, and say what you'd check first when a report's totals drifted. It also tests evaluation judgment: whether you can hold a consistent standard across ten similar outputs instead of grading on vibe, and whether you write critiques a model trainer could act on.

Logistics

  • Contractor engagement, US-based remote, no relocation or onsite component.
  • Asynchronous work delivered through a task platform; most contributors work in self-scheduled blocks rather than fixed hours.
  • Volume is project-driven and can fluctuate — treat this as supplemental or part-time unless told otherwise.
  • Observed pay band is $60–120/hr, varying by task complexity and calibration performance; nothing here is a guaranteed rate or a guaranteed volume.
  • No prior AI or ML background is expected. Domain fluency plus clear written reasoning is the whole requirement.