See what sovereign AI & data actually feels like
Most “sovereign AI” talk stays on slides. This cohort doesn’t. You work inside my infrastructure — the same Scaleway (France), Nebius (Finland) and self-hosted (Germany / Netherlands) setup I run for clients — and build something real with it.
No accounts to create, no API keys to buy, no cloud bill: you log into a preconfigured Uplevate environment and start.
What you’ll do
- Code with Mistral Vibe, preconfigured. Your Uplevate environment ships with Mistral Vibe and the Mistral Vibe VS Code extension already wired to my sovereign AI gateway — open it and go.
- Use real EU models through my gateway. Reach Scaleway models (Mistral Medium, gpt-oss-120b, Qwen3 / Devstral / Gemma / Pixtral / Voxtral) and Nebius models (DeepSeek, Kimi, GLM, Hermes, MiniMax, Nemotron) through my metered gateway — on my keys. A free local TurboQuant-accelerated llama.cpp model is there too.
- Search sovereignly. Web search runs through Staan, a European sovereign search index — no data leaves the EU.
- Own your data layer. Build and query a sovereign lakehouse on Apache Iceberg and DuckLake using my open tooling (DuckIceLake, Lakesh) — and learn the skills to manage it yourself.
Who it’s for
Engineers, data and platform teams, and technical leaders who need to understand — hands-on — what EU-sovereign AI and data infrastructure can and can’t do before committing to it.
Format
- Hands-on, cohort-based, small groups for real attention
- Everything runs live on Fineupp’s sovereign stack — nothing to install
- Dates, price and exact duration are being finalised — join the waitlist below and you’ll be first to know when the next cohort opens.
Your instructor
Led by Kevin Keller — a security and data advisor whose experience spans Oracle, IBM, Snowflake, Akamai and Sopra Steria — who designs and runs this exact sovereign stack day to day.