What the work involves

You will read AI-generated industrial engineering artifacts and judge whether they would survive contact with a real plant. That means checking whether a proposed line balance actually respects takt time, whether a stated Cpk follows from the data given, whether a kanban sizing calculation accounts for replenishment lead time and demand variability, and whether a cost analysis has quietly confused fixed and variable overhead. Much of the day is head-to-head comparison: two model responses to the same prompt, and you decide which is stronger on engineering rigor, process efficiency, and quality-control soundness — then write the justification that makes your ranking usable as training signal.

The written rationale is the deliverable. "Response B is better" is worth nothing; "Response B is better because it sequences the changeover using SMED principles and correctly treats the 40-minute internal setup as the binding constraint, while Response A assumes setup time is negligible and therefore understates WIP by roughly one shift" is the thing being paid for. Expect scenarios spanning production sequencing and scheduling, capacity and cost analysis, quality standards and control plans, inventory policy, and logistics coordination.

What the screen looks for

  • Four or more years in an industrial, process, quality, continuous improvement, or operations engineering role — with specifics: what you were optimizing, on what line or in what facility, and what changed measurably.
  • Fluency with the standard toolkit under follow-up questioning: SPC, DOE, capability indices, value stream mapping, line balancing, queueing and simulation, standard costing, MRP and inventory models.
  • Whether you can catch a plausible-sounding but wrong answer. Screeners frequently present flawed reasoning and watch whether you accept it.
  • Clear written English — the rationales are the product, and thin or vague writing fails the task even when the judgment behind it is correct.

Logistics

Fully remote and largely asynchronous; tasks are pulled from a queue rather than scheduled. The commitment is 20+ hours per week over an engagement of roughly four to six weeks, starting immediately. Payment is task-based at the outset and released on first task approval; experts who clear that first task inside the required window move to hourly for the rest of the project. The $70–110/hr band reflects rates observed on comparable Mercor engineering projects and is not a guarantee — actual rates depend on the project, your assessed depth, and the platform's own calibration.