AI Research Fellowship

Sovrano AI's Expert Track is built for graduate students whose academic depth creates genuine value in AI post-training. This is an ongoing, rolling opportunity - you work when projects are available, at the pace that fits your studies. Open to Master's students and graduates anywhere in Europe.

Master's Fellows on this track contribute directly to evaluation pipelines commissioned by the world's top LLM companies. Your work is not abstract - it directly influences how models are fine-tuned by the labs building GPT, Claude, Gemini, and the next generation of AI systems. That is an unusual thing to be able to say as a student.

Output quality is tracked at the task level. Fellows who produce inconsistent or low-quality evaluations are removed from projects without notice. We value reliability and precision above all else.

KEY RESPONSIBILITIES Produce structured, domain-grounded evaluations of LLM outputs on a rolling project basis Write ideal completions and reasoning chains in your subject area Evaluate model behaviour across safety, factuality, and domain accuracy dimensions Contribute to red-teaming and adversarial stress-testing tasks as needed IDEAL QUALIFICATIONS Master's students or graduates from a Sovrano AI partner institution anywhere in Europe Domain expertise in STEM, social sciences, law, medicine, or linguistics preferred High self-discipline - async work requires you to manage your own output without hand-holding Proven ability to produce high-quality written work under guidelines WHAT SUCCESS LOOKS LIKE A direct working relationship with the post-training operations of the world's leading AI labs Ongoing supplemental income at €20–28/hr, paid weekly Continuous exposure to state-of-the-art model development practices A growing track record as a verified Sovrano AI Expert Fellow Priority access to full-time and senior fellowship opportunities as the network expands CONTRACT & PAYMENT TERMS

No minimum commitment. Project-matched on a rolling basis. €20–28/hr depending on task complexity and domain. Paid weekly.