Physics Expert (PhD/Master's)

What you'll do

This is the physics-specific posting among Turing's expert problem-writing roles. You design and solve physics problems hard enough to expose a language model's limits, and write each solution as clear, step-by-step reasoning. You work with LLM researchers so problems serve a stated evaluation goal, concentrating on places where models stumble: abstraction, multi-step reasoning, symbolic manipulation. You also help define benchmarks built on physics curricula from early undergraduate to PhD level.

The listed skill areas are classical mechanics, electromagnetism and optics, and thermodynamics and statistical physics, which gives a good idea of the syllabus you would be writing against.

Who fits

• A strong physics foundation, at the level of engineering entrance exams and graduate or PhD programs
• Candidates currently pursuing a Master's, PhD or postdoc in physics or applied physics are eligible and encouraged to apply (a preferred qualification)
• Good research, analytical and feedback skills, plus creative thinking
• The ability to explain physics clearly in simple language and visuals
• A desktop or laptop and good internet, and the discipline to work independently

What it pays

The ad does not publish a rate, hours or contract length. Turing sets rates per project; see Turing.

Worth knowing

Good:

• Clearly physics-only, so you know the subject before applying
• Students and researchers still in training can apply
• Possible extension based on performance and project needs

Less good:

• No pay figure, hours or duration
• Contractor assignment with no medical or paid leave
• Open since May 2025 per the board date
• The ad does not describe assessments; Turing's help centre mentions skill tests and, for some roles, a live interview

About this listing

Posted on Turing's job board on 20 May 2025 and read on 18 October 2026. Requirements and perks are Turing's own text. A broader STEM version of this role, covering more than physics, is also on the board. See Turing.