What the work actually looks like

You are handed model outputs on hardware problems and asked to say whether they are right, where exactly they go wrong, and what a correct answer would have been. In practice that means things like: checking a Verilog or SystemVerilog module for latch inference, blocking/non-blocking misuse, or reset-domain errors; auditing a timing-closure explanation for whether setup and hold slack are reasoned about correctly; judging whether a proposed PCB stackup, return-path, or decoupling strategy would survive a real EMC test; verifying that an I2C/SPI/UART or DDR interface description matches the actual protocol; or reviewing power-integrity, ESD, thermal, and DFM claims. Some tasks are ranking two candidate answers, some are authoring a hard problem plus a reference solution, some are rubric-based scoring where you have to justify a score in writing.

  • Written rationales matter as much as the verdict — "wrong" without a cited reason is a rejected task
  • Tasks arrive in batches; expect calibration rounds and re-review of your early work
  • Specialisation is fine and often preferred: digital design, analog/mixed-signal, RF, embedded firmware, or semiconductor process

What the screen is looking for

micro1 runs an AI interviewer that asks about your background and then pushes into technical follow-ups on whatever you claim. It is testing whether your hardware knowledge is first-hand — whether you have taped out, brought up a board, debugged on a scope or logic analyser, closed timing, or shipped a product — and whether you can articulate a subtle error clearly and quickly under a follow-up question. Vague seniority claims collapse fast; a specific bug you found and how you found it holds up. There is also a plain availability check: hours per week, timezone overlap, and whether you can commit for the length of a project.

Logistics

Fully remote and largely asynchronous, with deadlines per batch rather than fixed shifts. Most contributors work 10–20 hours a week alongside a day job; some projects offer more volume, and volume is genuinely uneven between engagements. Rates in the $50–100/hr band are what contributors have reported for this category — the figure attached to any given project depends on the domain, the difficulty tier, and how you calibrate, and nothing here is a guarantee.