Remote | Physics Expert; PhD​/Master's

Selected professionals will design and solve complex physics problems, develop clear step-by-step explanations, identify areas where advanced AI systems struggle, and collaborate with research teams on evaluation strategies and benchmarks spanning early undergraduate through doctoral-level physics.

• Design challenging physics problems across advanced subject areas
• Create tasks that probe limitations in scientific reasoning
• Develop problems requiring abstraction and multi-step analysis
• Incorporate symbolic manipulation and quantitative reasoning where appropriate
• Maintain suitable difficulty across undergraduate through PhD-level topics

• Solve complex physics problems accurately and systematically
• Produce clear and rigorous step-by-step solutions
• Explain assumptions, calculations, and intermediate reasoning explicitly
• Apply structured physical reasoning to challenging scenarios
• Present conclusions in a format suitable for evaluation and review

• Evaluate how language models perform on physics-focused tasks
• Identify recurring weaknesses in conceptual understanding
• Analyse errors involving abstraction, symbolic manipulation, and multi-step reasoning
• Distinguish conceptual mistakes from calculation or interpretation errors
• Provide structured observations to support model improvement

• Contribute to new physics evaluation benchmarks
• Align problem design with specific research objectives
• Develop tasks spanning different levels of technical complexity
• Refine evaluation criteria based on observed model behaviour
• Support consistent assessment across advanced physics domains

• Collaborate remotely with research and technical teams
• Explain complex physics concepts in clear and accessible language
• Provide constructive feedback and detailed annotations
• Communicate analytical findings in a structured format
• Work independently while maintaining consistent project quality

• Currently pursuing or holding a Master's, Ph.D., or postdoctoral position in Physics, Applied Physics, or a related field
• Strong foundation in advanced physics at graduate or doctoral level
• Expertise may include quantum mechanics and quantum information
• Background in relativity is relevant
• Nuclear and particle physics expertise is valuable
• Astrophysics and cosmology backgrounds are welcome
• Condensed matter physics expertise is relevant
• Strong knowledge of classical mechanics
• Strong understanding of electromagnetism and optics
• Familiarity with thermodynamics and statistical physics
• Strong analytical and research capabilities
• Ability to solve complex physics problems using a structured and logical approach
• Comfortable with abstraction, symbolic manipulation, and multi-step reasoning
• Ability to explain physics concepts clearly using simple language, visuals, and physical intuition
• Strong critical-thinking and creative problem-solving skills
• Ability to provide detailed annotations and constructive feedback
• Excellent written English communication
• Strong remote collaboration skills
• Self-motivated and able to work independently
• Reliable desktop or laptop setup with a stable internet connection

• Contractor assignment / freelance engagement
• Fully remote
• Source materials describe the opportunity as full-time, while the formal engagement type is contractor/freelancer
• Work will involve physics problem design, solution development, model evaluation, and benchmark creation
• Projects may span quantum mechanics, relativity, nuclear physics, particle physics, astrophysics, cosmology, condensed matter, classical mechanics, electromagnetism, optics, thermodynamics, and statistical physics
• Evaluation may cover topics ranging from early undergraduate through PhD level
• Collaboration will involve research teams working on advanced language-model evaluation
• Contract extension may be available depending on performance and project requirements
• No medical or paid-leave benefits are included under the stated contractor arrangement
• Work must be completed without using confidential, proprietary, unpublished research, restricted datasets, or otherwise…