Digital Sovereignty & Data Ownership

Your data.
Your systems.
Your rules.

I plan IT landscapes that belong to your organisation and guide the way there: open standards, operated in your own house, hardened. Two questions, one answer — who owns your data, and who can reach it.

25+

Years of systems engineering

100%

Vendor-neutral, no commissions

DE·EN

Advice in German and English

A bunch of keys resting on a small home server
Inside your house
  • Cloud and AI applications
  • Documents and knowledge base
  • Backups and version history
  • Access rights and audit logs

None of it sits with an outside provider.

Services

Four routes back to control

From assessment to running system — vendor-neutral, on open standards, operated in your own house.

Sovereignty check

We establish where your data actually sits, who has lawful access to it, and which dependencies would leave you unable to act in a crisis. The result is a prioritised order: what to change first, what later, and what to leave alone.

Out of someone else’s cloud

A staged move back to open source and systems you control: recover your data, cut licence costs, secure your ability to operate, with no big bang and no interruption to daily business.

AI in a cage

Language models and assistants working on your own documents, run locally, so your knowledge never lands with an outside provider. Here AI is a tool, not the point: it should use your data ownership, not break it.

Hardening & defence

Close the attack surface before it is found: separate access, switch off unused services, automate updates, rehearse recovery. I have run servers across many providers for over twenty years, including the incidents you learn from.

Security

Attacks are automated. Are your defences?

What used to be a targeted attack is now production-line work: automated systems probe every reachable address for every known weakness, around the clock. You are not attacked because you are interesting — you are attacked because you are reachable. For a small organisation that is existential: one encrypted server and the business stops.

  1. Shrink the attack surface instead of buying products What is not running cannot be attacked. The most effective measures carry no licence fee — they cost decisions.
  2. Backups that actually restore A backup only becomes one once the restore has been rehearsed. Anything else is hope that consumes disk space.
  3. Access that makes no single person a risk Separated rights, traceable changes, and no shared administrator password sitting in a chat history.

Test questions

Sovereignty with an asterisk

Now that large providers call their data centres “sovereign”, the practice has a name: sovereignty washing. Four questions that separate a promise from a fact — including when you ask them of me.

  1. Who holds administrative access — and under which jurisdiction? A European data centre is not enough if the parent company answers to a foreign legal system.
  2. Can you inspect the source code? What cannot be verified has to be believed. Open source turns trust into something you can check.
  3. Can you change provider without losing your data? Open formats and documented interfaces decide whether a decision stays reversible.
  4. Does your system keep running if the provider shuts down? A system is sovereign only when its continued operation does not depend on a contract staying alive.

Projects

What I am working on

Current work where sovereign systems are not theory but the build specification.

In-house retrieval systems

Knowledge systems that answer questions from an organisation's own material. Embeddings and index run locally; the knowledge base never leaves the organisation. Currently in use at a cooperative.

Agentic AI integration

Assistants that do not merely answer but act — with defined rights, tested boundaries and traceable access. Here the frame is the actual work.

EUCUBE

Platform work in a European and cooperative context — where independence from large providers is a precondition of the undertaking.

InvestTracker

An application for evaluating and tracking investment and financial data. Sensitive figures, own infrastructure, nothing handed to third parties.

In your own house

Where your systems run

Models, index and logs on hardware that belongs to you. Not in someone else’s cloud.

Compact computer with a network cable plugged in
Smartphone on a dark desk
Computer with code running on screen

Illustration. The hardware shown stands symbolically for locally operated systems. Images generated locally with AI.

Zsolt Thomas Szentirmai
25+ years
systems engineering

About

Zsolt Thomas Szentirmai

Dipl.-Wirtsch.-Ing. · Systems engineer for digital sovereignty

I have been responsible for digital systems since 1999: as project manager for European change management at Toshiba Electronics Europe, as managing director of a federally funded eBusiness centre at the Munich Chamber of Commerce, and since 2008 with my studio Streampark Media Networks in Berlin.

Today I work on one question: who owns the systems an organisation depends on? I plan IT on open standards, guide the delivery and review the result. Artificial intelligence belongs in that picture, but as a tool — run locally, so it uses your data ownership rather than surrendering it.

What I recommend, I know from operating it. As senior consultant and founding member of Platform Coops Germany eG, I work in particular with cooperative and public-benefit organisations — where independence from large platforms is not merely an IT question but one written into the statutes.

Dipl.-Wirtsch.-Ing., industrial engineering Senior consultant, Platform Coops Germany eG Founder, Streampark Media Networks German · English
View LinkedIn profile

Contact

Let’s talk.

Whether it is an assessment, an exit from a dependency, or a system that has to stay in-house — write and tell me what you have in mind.

Map: Neue Plantage 6, Potsdam
© OpenStreetMap contributors