# Sovrano AI

> Europe's expert-grade, GDPR-native human evaluation platform for AI. We connect Europe's best young talent with AI companies building the future.

Sovrano AI is the domain-expert talent layer for the teams training, evaluating, and red-teaming AI. We staff your RLHF, red-team, SFT authoring, multilingual evaluation, and reasoning-benchmark work with European evaluators you plug into your own pipelines. Our pool is 120,000+ verified evaluators across 100+ European universities, GDPR-native and EU AI Act ready. All work is done by people, not pipelines, which is what makes it usable for frontier-model training.

## For AI Companies
- [For AI Companies](https://sovrano.ai/for-ai-companies): Domain-expert European evaluators staffing your RLHF, red-teaming, and evaluation work. University-verified talent across 23 languages.

## For Students
- [For Students](https://sovrano.ai/for-students): Get paid to make AI smarter. Flexible, remote AI evaluation work matched to your domain of expertise.
- [RLHF Guide](https://sovrano.ai/for-students/rlhf): What reinforcement learning from human feedback actually looks like as a job. Hands-on explainer for evaluators.
- [Deep Annotation Guide](https://sovrano.ai/for-students/deep-annotation): How domain experts annotate complex documents for AI training.
- [Red-Teaming Guide](https://sovrano.ai/for-students/red-teaming): How to stress-test AI safety by finding edge cases where models fail.
- [Referral Program](https://sovrano.ai/referral): Refer an evaluator and earn 20% of everything they make on Sovrano AI.

## For Universities
- [For Universities](https://sovrano.ai/for-universities): Partner to give your students hands-on AI experience. Structured, credit-eligible internship that fits your calendar.

## About
- [About Sovrano AI](https://sovrano.ai/about): The team, the mission, and why Europe is building the human evaluation layer for AI.
- [Careers](https://sovrano.ai/careers): Open roles in Barcelona. Each one owns a part of the company. Apply at careers@sovrano.ai.

## Blog and Help
- [Blog](https://sovrano.ai/blog): Perspectives on RLHF, AI evaluation, and how the industry is evolving.
- [Help Center](https://sovrano.ai/support): Answers for students, AI companies, and universities — account help, task guidance, payments.

## Key facts
- Headquarters: Barcelona, Spain
- Coverage: 23 European languages
- Workforce: 120,000+ verified evaluators
- University partners: 100+ European institutions, EFMD-certified
- Compliance: GDPR-native, EU AI Act ready
- Payment: Weekly via Deel (regulated global payroll)
- Contact: hello@sovrano.ai
- LinkedIn: https://www.linkedin.com/company/sovrano-ai/
- YouTube: https://www.youtube.com/@SovranoAI

## Task archetypes (what evaluators do)
- **RLHF**: Pairwise comparison of model outputs to teach reward models what better means in specialised domains.
- **Deep annotation**: Marking up complex documents (legal, medical, financial) with structured labels so models learn domain nuance.
- **Red-teaming**: Designing adversarial prompts to surface where models hallucinate, break safety, or reason poorly.
- **Multilingual evaluation**: Native-language scoring across 23 European languages, including regional variants.

## Who works here
- Domain experts (finance, legal, consulting, medicine, engineering) who moonlight as evaluators.
- MBA and graduate students at European business schools and universities, often through structured for-credit internships.
- Early-career researchers specialising in AI safety and alignment.