Robotics Internship (Work From Home)

About the internship

This remote internship is aimed at highly capable scientific and quantitative thinkers who can help build advanced evaluation data for frontier AI systems. The work focuses on creating demanding, text-only reasoning challenges that can test the limits of next-generation models.

The role is far beyond routine data entry or typical content writing. It calls for precise, high-difficulty problem design that can be checked automatically for correctness.

What you will do

• Build sophisticated STEM problem sets and evaluation datasets in text and mathematical form only.
• Write clearly specified problems whose answers can be verified through code or other programmatic checks.
• Develop tasks that meaningfully stress-test advanced AI models while maintaining a very high difficulty bar.
• Create challenging questions across mathematics, theoretical physics, chemistry, and life sciences such as molecular biology and biotechnology.
• Support AI alignment research, model benchmarking, and automated verification workflows.

Who can apply

Applicants must be available for a work-from-home internship, be able to begin between 30 June 2026 and 4 August 2026, and commit to 6 weeks. Candidates should also have relevant skills and interest in the work, and must be Mechanical Engineering students.

Domain and academic expectations

• Mathematics should account for roughly one-third of the work, with emphasis on advanced theory, proofs, and algorithmic structures.
• The remaining work is expected to span advanced STEM areas, especially theoretical physics, organic and inorganic chemistry, and life sciences including molecular biology and biotechnology.
• Strong preference is given to PhD candidates or PhD graduates in Mathematics, Physics, Chemistry, or Biological Sciences from Tier-1 Indian institutes such as IITs, IISc, IISERs, or leading central universities.
• Final-year master’s students in M.Sc. or M.Tech. programs from top institutions with an exceptional research record may also be considered.
• Candidates should have a deep interest in AI alignment, model benchmarking, and automated code-based verification systems.

Compensation

The listed stipend is USD 30,000 to 100,000 per month. An additional incentive component of USD 10,000 to 50,000 per month is also mentioned.

Perks and working style

• Flexible work hours
• Five-day work week
• Fully remote setup
• Tier-based variable compensation linked to subject expertise and evaluation quality
• Direct exposure to the training and evaluation process for enterprise-grade large language models

Additional details

This internship requires original, highly complex evaluation logic in the chosen subject area as part of the early screening process. Applicants should be prepared to share a brief example of their research background or primary area of expertise.