What the work actually involves
You write problems from designs you have actually built: the specification, the constraints, the operating conditions, and the reference answer the model's attempt is graded against. Then you grade — whether the topology fits the requirement, whether noise and stability margins hold, whether timing closes across corners. A recurring task is catching designs that simulate cleanly and would fail on a real board: parasitic inductance in a return path, tolerance stack-up that eats the margin, a ground scheme that couples switching noise into a sense line. Written justification is the deliverable; a verdict without reasoning is not usable as training signal.
What the platform screens for
- Bench experience, not coursework. The listing says it plainly: where you trained matters less than whether you have debugged your own design and found what the simulation missed.
- Specificity under follow-up. The AI interview will push on a design you claim — part numbers, the measurement, the failure mode, what you changed.
- Calibration. Can you separate a design that is wrong from one that is merely not how you would have done it, and say which is which in writing.
- Honesty about ambiguity. Flagging an underspecified task is rewarded; guessing the intent and proceeding is not.
Logistics
This is a standing listing, not an opening. Applying puts you in the electrical engineering pool after background verification; you will never receive an accept or reject on the listing itself. Matching can happen within a week or take several months, depending on what clients are running. Individual project listings name the rate, the hours, and the client, and the hiring decision is made there.
Work is remote and asynchronous, typically part-time alongside a day job, with deadlines per batch rather than fixed hours. Application is a resume, a location confirmation, and roughly a 20-minute AI interview.