What the work involves

You spend most of your time judging whether an AI model's answer would actually survive contact with Fusion. That means reading a prompt ("build a parametric bracket that adapts to three hole spacings," "why is this loft failing," "generate a 2.5-axis adaptive clearing strategy for this pocket"), reproducing the model's proposed workflow in Fusion, and writing down precisely where it goes wrong — a mis-ordered timeline, a sketch that isn't fully constrained, a joint that should have been a rigid group, a feed and speed that would snap the tool.

Typical task types include:

  • Response comparison — ranking two or more AI answers on technical correctness, workflow efficiency, and whether a working engineer could follow them.
  • Error annotation — flagging hallucinated commands, non-existent menu paths, deprecated UI references, and API calls that don't exist in the Fusion Python/JavaScript API.
  • Gold-standard authoring — writing the correct answer yourself, including step sequences, parameter tables, or corrected code.
  • Prompt creation — constructing realistic, hard tasks drawn from actual design or machining work you've done.

What the platform screens for

micro1 runs an AI-led interview before human review. It probes for depth you can only have from real hours in the software: how you'd restructure a timeline to make a model robust to parameter changes, when you'd use a surface workflow instead of solids, how you'd fix a Fusion assembly that regenerates unpredictably, what you check before posting a CAM operation. Expect follow-ups that go one or two levels past your first answer, and expect vague claims to be tested. Manufacturing context — CNC, sheet metal, 3D printing, injection molding DFM — carries significant weight, as does experience with the Fusion API, generative design, or simulation.

Logistics

Remote and asynchronous. Most contributors work 10–20 hours per week against rolling deadlines, though volume fluctuates with project cycles and there are quiet stretches between engagements. You'll generally need your own Fusion license (personal, startup, or commercial) and a machine that can run it comfortably. Pay is hourly and set per project; the $50–120 range reflects observed rates rather than a guarantee, with the upper end tied to API fluency, CAM expertise, or credentialed engineering backgrounds.