What the work actually is
You will be watching short clips — often repetitive, often unglamorous — of robotic arms attempting tasks: picking up a cup, opening a drawer, stacking blocks, folding cloth. For each session you apply a grading rubric supplied by the customer: did the task complete, at what point did it fail, was the grasp secure or did the object slip, did the arm collide with anything, was there an occlusion that makes the clip ungradable. The annotation itself is done in the customer's tooling; the skill being bought is your eye. Filmmakers and cinematographers are targeted specifically because the job is sustained frame-by-frame observation — noticing the half-second wobble, the shadow that reveals contact, the moment the object left the gripper — and the discipline of scrubbing back rather than guessing from the outcome shot.
What the screen looks for
micro1's process is AI-led: a conversational interview, usually 20–40 minutes, followed by task-based calibration work. The screen is checking three things. First, that your production experience is real and specific — formats, tools, roles, footage volumes you've actually reviewed. Second, that you can reason about ambiguity rather than dodge it: what you do when the rubric doesn't cover the clip in front of you, how you decide between "failed" and "unscoreable." Third, whether you'll apply a guideline you personally disagree with, which is the most common reason experienced creatives wash out of annotation work. Expect follow-ups that press on any vague answer.
Logistics
- Fully remote, contractor engagement, asynchronous within batch deadlines
- Typically part-time and project-based; hours often flexible but batches carry turnaround windows
- Requires a reliable machine, stable connection, and a display good enough for frame-accurate scrubbing
- Written English strong enough to document annotation rationale and flag guideline gaps
- Volume fluctuates with customer demand; treat it as supplemental rather than a fixed schedule