AI Company Pays Scientists $15K/Month for Research Fellowship

Scientists and engineers now have a chance to shape the future of AI-powered discovery while earning nearly $50,000 for three months of work. Anthropic, a San Francisco AI safety company, just opened applications for its STEM Fellows Program, bringing domain experts together with cutting-edge AI tools to tackle real scientific challenges.

The fellowship runs full-time from mid-June through mid-September in San Francisco. Fellows receive approximately $3,800 per week, access to advanced AI models like Claude, and direct mentorship from Anthropic's research team.

The program focuses on building what Anthropic calls an "AI scientist," a system capable of reasoning through complex experiments and making genuine scientific discoveries across disciplines. Fellows work on defined projects that match their expertise, from improving AI reasoning in materials science to integrating climate modeling tools into research workflows.

Participants spend their time designing evaluation frameworks to test AI capabilities, identifying where models fail or produce incorrect results, and creating datasets that help AI systems better understand scientific problems. The work is practical and scoped to deliver real results within the three-month window.

Anthropic wants PhDs in STEM fields or candidates with equivalent research experience. Strong analytical skills and scientific reasoning matter more than machine learning expertise. The company specifically encourages applications from scientists interested in how AI can speed up discovery, even if they've never worked directly with AI tools before.

The application process includes an online submission, technical assessment, take-home exercise, and discussion with a mentor about research interests. The deadline is May 15, 2026, with decisions expected by June 1.

The Ripple Effect

This fellowship represents a shift in how AI companies are approaching scientific advancement. By paying competitive stipends and bringing in domain experts rather than just AI specialists, Anthropic is acknowledging that building useful scientific AI requires deep knowledge of actual science, not just better algorithms.

The program also opens doors for researchers who might not otherwise have access to frontier AI systems. Fellows get hands-on experience with tools that could transform their home fields, then return to their institutions with new skills and perspectives.

Anthropic emphasizes that diversity drives better AI development and encourages applications from underrepresented groups. The company even addresses imposter syndrome directly, urging qualified candidates not to count themselves out.

For scientists curious about AI's potential to accelerate their work, this summer could be both lucrative and transformative.