Mathematics Internship (Work From Home, Part Time)

About the Internship

This internship is focused on building advanced evaluation data for cutting-edge AI systems. The work is intended for exceptionally strong scientific and mathematical candidates who can create difficult, text-based reasoning tasks that challenge next-generation models.

The role is not centered on routine data entry or conventional content creation. Instead, it calls for sophisticated problem design and solution verification, with outputs that can be checked programmatically for clear correctness.

What You Will Do

• Develop demanding STEM problems and evaluation datasets in text and mathematical formats, with no multimodal content involved.
• Build rigorously structured questions whose answers can be validated definitively through computer code.
• Create high-difficulty assessment tasks that push and test the limits of advanced AI models.
• Maintain an elite difficulty threshold, including tasks where models should fail to answer correctly at least half the time across multiple attempts using the Pass@8 benchmark.

Who Can Apply

This opportunity is open only to candidates who can work remotely, are available for a 2-month internship, and can begin between 30 Jun 2026 and 4 Aug 2026. Applicants should have relevant skills and a strong interest in the subject area.

Strict Candidate Profile

• PhD scholars or PhD candidates in Mathematics, Physics, Chemistry, or Biological Sciences from top-tier Indian institutions such as IITs, IISc, IISERs, and leading central universities.
• Final-year master's students pursuing M.Sc. or M.Tech from premier institutions with an outstanding academic record in core research may also be considered.
• Candidates should have a strong interest in AI alignment, model benchmarking, and code-based verification systems.

Domain Focus

• Mathematics contributes around 33% of the project scope and includes advanced theory, proofs, and algorithmic structures.
• The remaining 67% covers other STEM areas, including advanced theoretical physics, advanced organic/inorganic chemistry, and life sciences such as molecular biology and biotechnology.

Compensation

The internship offers a stipend of USD 15,000 to 30,000 per month. In addition, there is incentive pay of USD 5,000 to 10,000 per month, with compensation varying based on domain expertise and evaluation quality.

Perks and Benefits

• Certificate on completion
• Letter of recommendation
• Flexible working hours
• Five-day work week
• Fully remote work arrangement
• Opportunity to contribute to the training and evaluation workflows of enterprise-grade LLMs

Additional Information

The organization works as an operating layer for practical AI, partnering with frontier AI labs, enterprises, and government agencies to help build and validate the next generation of intelligent systems. Its evaluation science platform uses model-agnostic methods together with a verified network of more than 10,000 domain experts across law, finance, science, medicine, engineering, public policy, and related areas to support structured human evaluations, red-team exercises, and alignment workflows that improve model quality and safety.

Application Notes

Applicants should be ready to share a brief explanation of their research background or main area of expertise and may be asked to demonstrate a short sample of highly complex, text-based evaluation logic during the initial screening.