Remote Electrical Engineer — AI Model Evaluator

DataAnnotation

Alabama

Remote

USD 55,000 - 172,000

Full time

10 hours ago

Be an early applicant

Application generator

Turn this role into an interview — a resume and cover letter built around what this employer wants.

Get past ATS filters

Job summary

DataAnnotation is seeking an experienced Electrical Engineer to help teach AI to reason about circuits, systems, and physical constraints. You will design challenging engineering problems and judge frontier AI outputs against professional standards.

You’ll evaluate analog, RF, mixed-signal, electromagnetics, and power systems, critique model derivations, and flag failures with concrete evidence. Remote, worldwide engagement with US-based projects.

Qualifications

Fluency in English (native or bilingual level).
2+ years of industry experience in circuit or electronics engineering.
Background in electrical engineering, electronics engineering, or a closely related discipline.
Strong command of circuit analysis and design tradeoffs, and the ability to defend component-level and system-level decisions.
The ability to articulate why an answer is wrong, not only that it is.
Working comfort with AI tools, or the ability to pick them up quickly.
Bachelor’s degree is preferred but not required.
Only applicants based in Australia, Canada, Ireland, New Zealand, the United Kingdom, or the United States will be considered for this role.

Responsibilities

Design challenging, realistic engineering problems from your own practice: a circuit to specify, a control loop to stabilize, a power stage to size.
Run those problems through frontier AI models and judge the output against your professional standard.
Compare two model versions on identical prompts and call which performed better, and where each fell short.
Critique model-generated derivations, schematics, component selections, and calculations, and write the reasoning behind your call.
Flag concrete failures with evidence: wrong assumptions, sign and unit errors, unphysical results, invalid topologies, fabricated datasheet values, or off-brief interpretation.