The frontier of AI capability is currently defined by one bottleneck: the availability of verified STEM experts who can produce and evaluate high-quality technical reasoning data. The shift from generalist annotation to expert-produced reasoning chains is the defining trend in AI post-training - and the world's top LLM companies are competing aggressively for exactly this expertise.
STEM Expert Fellows work directly on technical training projects commissioned by the world's leading AI labs. The companies building the most capable models on the planet - the systems that will define how AI handles mathematics, physics, engineering, and science for the next decade - need people like you to make them work. This is not a simulation. This is the actual work, and it is consequential.
This is the most demanding role in the fellowship. You will be expected to produce publication-quality reasoning traces, identify subtle technical errors that generalist reviewers miss entirely, and maintain output quality under extended, intensive task cycles. If you are not comfortable working through difficult technical problems alone for hours at a time, this is not the right role for you.
KEY RESPONSIBILITIES Produce step-by-step solution traces for hard technical problems in your domain - mathematics, physics, computer science, engineering, biology, chemistry Evaluate AI-generated technical solutions for correctness, completeness, and reasoning quality Write ideal model responses to complex STEM prompts at a level that would satisfy a domain expert reviewer Identify and classify failure modes in model reasoning: calculation errors, conceptual misunderstandings, unjustified leaps Contribute to benchmark creation and evaluation framework design for STEM domains IDEAL QUALIFICATIONS Master's students, PhD candidates, or recent STEM graduates from any Sovrano AI partner university in Europe Strong academic record in mathematics, physics, CS, engineering, biology, or chemistry Exceptional ability to produce clear, precise, step-by-step written solutions High self-discipline and tolerance for extended solo focus work Genuine interest in AI development - not just the paycheck WHAT SUCCESS LOOKS LIKE A direct working relationship with the technical training operations of the world's top LLM companies - an increasingly rare and valuable credential in any technical career €28–35/hr - the highest pay rate in the Sovrano AI Fellowship, paid weekly Verified Sovrano AI STEM Expert Fellow credential, recognised by leading AI labs Deep insider knowledge of how frontier models are trained on technical reasoning tasks Priority access to advanced research collaboration opportunities as Sovrano AI scales CONTRACT & PAYMENT TERMS
Ongoing, rolling availability. STEM expert slots are limited and allocated based on qualification and demonstrated quality. €28–35/hr depending on domain and task complexity. Paid weekly. May be compatible with academic supervisory arrangements - discuss with your institution.