AI Tutor, Physics Specialist (contract), Handshake AI

Handshake · Enterprise · Remote · AI Trainer

This role focuses on evaluating AI models, specifically in physics, by crafting and assessing challenging problems, probing model reasoning, and identifying failures using adversarial prompting. It involves providing expert critique of AI responses and ensuring quality benchmarks are met. Prior experience in AI data annotation or RLHF is required, with a PhD in Physics or a related field.

What you'd actually do

  1. Develop and evaluate high-difficulty physics prompts spanning classical mechanics, quantum mechanics, electromagnetism, thermodynamics, and related areas
  2. Use adversarial prompting techniques to expose errors in model reasoning and problem-solving
  3. Provide expert critique of AI-generated responses, assessing both correctness and depth
  4. Work closely with project leads to uphold quality benchmarks

Skills

Required

  • PhD in Physics or a closely related field
  • Graduate-level expertise across multiple areas of physics
  • Prior hands-on experience in AI data annotation or RLHF
  • Excellent written communication
  • Analytical skills

Nice to have

  • Publications in peer-reviewed chemistry or physics journals
  • Experience with adversarial prompting
  • Model evaluation experience
  • AI red teaming experience
  • Teaching experience in physical chemistry or theoretical sciences
  • Tutoring experience in physical chemistry or theoretical sciences
  • Curriculum development experience in physical chemistry or theoretical sciences
  • Experience with computational chemistry tools (e.g., Gaussian, ORCA, MATLAB)
  • Background in scientific annotation
  • Background in technical quality assurance

What the JD emphasized

  • AI model evaluation
  • probe model reasoning
  • identify where models fail
  • adversarial prompting
  • expert critique
  • quality benchmarks

Other signals

  • AI model evaluation
  • probing model reasoning
  • identifying model failures
  • adversarial prompting
  • expert critique of AI-generated responses