Nuclear Engineering & Safeguards Experts for Red Team

In short

This remote freelance role with Mercor has 905 openings. The team is building a panel of nuclear experts to test how AI models handle sensitive technical questions. You would help check that models answer legitimate questions fully while refusing dangerous ones. It may suit you if you know nuclear fuel, safeguards or nonproliferation and enjoy careful writing.

What you will do

• Write tough single-turn prompts in your field, sorted into benign, dual-use and adversarial levels.
• Check the model's replies against a set policy standard and decide whether each one was handled correctly.
• Write a reference answer that shows what a correct reply looks like, with the technical reasoning behind it.
• Explain each judgment in writing so that a non-specialist can follow your logic.

What you need to apply

• Depth in the nuclear fuel cycle, plus some experience with how nuclear material is controlled, accounted for or diverted.
• Safeguards or nonproliferation experience, which is the strongest signal for this role.
• Strong technical writing skills. A sample, published research or a link to past work is welcome.
• Ability to read and write about misuse scenarios for long stretches, with breaks allowed whenever you need them.
• No use of classified, export-controlled or NDA-covered information in your work.

Who can apply

Open to people anywhere in the world, as no country restriction is stated. Relevant backgrounds include safeguards analysts, nuclear engineers, nuclear security engineers, nuclear forensics scientists and nonproliferation analysts.