What the work involves
You review and score AI-generated responses to scenarios drawn from your day job: a caller reporting smoke on the second floor, a structure fire with conflicting address information, an unresponsive patient while EMS is still eight minutes out. Tasks typically include grading model transcripts of simulated 911 calls, writing reference answers that show correct call-taking sequence and pre-arrival instructions, flagging responses that skip location verification or give unsafe medical direction, and comparing two model outputs to explain which one a real dispatcher could act on.
Some projects lean toward radio and incident-management realism — apparatus assignments, ICS terminology, mutual aid, MAYDAY handling, NIMS resource typing. Others focus on the caller-facing side: EMD/EFD protocol adherence, panic control, hysterical caller techniques, and knowing when to stop asking questions and start dispatching. Written justification matters as much as the score; reviewers want to see the specific protocol step or standard you're citing.
What the platform screens for
micro1 runs an AI-led interview before you reach paid work. Expect follow-ups that press on specifics: which protocol system your center uses, how you'd sequence questions on a particular call type, what a given radio phrase means in your jurisdiction. Vague or generalized answers rarely survive the second follow-up. Screening also weighs whether you can articulate why an answer is wrong, not just that it is — that's the core skill in evaluation work.
Logistics
- Fully remote and asynchronous; you claim tasks from a queue.
- Typical commitments run 5–20 hours per week, though volume fluctuates with project cycles.
- Compatible with shift work — no live meetings required on most projects.
- Pay is stated as observed on the platform for this category and is not guaranteed; rates vary by project and calibration performance.