About
Technology, business, and the systems where they meet.
I’ve always been at the intersection of technology and business. Now I build the systems where they meet.
My path moved from electronics into international business consultancy, FinTech product ownership, enterprise delivery, and now I’m exploring the frontiers of AI systems architecture.
That arc matters.
I don’t approach AI as a feature layer. I approach it as:
- A Delivery Problem
- A Governance Problem
- An Organisational Design Problem
In practice, that means building systems that:
- Scale under real conditions
- Hold Decisions over time
- Survive Contact with Reality
Professionally
I work as an IT Delivery Manager in Vienna at the digital arm of a global organisation, operating in multiple jurisdictions and regulated markets.
25+ subsidiaries. 1,500+ employees.
I lead group-wide initiatives across:
- Aligning IT Infrastructure Standards across the subsidiaries
- Improving Digital Security Posture
- Leading AI Rollout Initiatives and collaborating with Microsoft in building AI capabilities
AI needs to be functional enough to deploy at scale, and controlled enough to trust with real decisions.
Building and testing independently
In parallel, I build and test systems independently to explore where current AI architectures hold and where they break.
I’m building DotOS, a reflexive AI operating system. 5D is the engine that governs its actions and keeps it safe.
Stack:
- LangGraph orchestration
- Local models via Ollama
- PostgreSQL with vector search
- Discord interface
This is not a product. It’s a working environment.
I use it to test:
- Autonomy vs Control
- Orchestration across agents
- Memory under Degradation
- Where Human Oversight is still required
I designed the 5D Risk Governance Model:
- Weighted decision engine
- Markov-based drift detection
- Real-time routing between autonomous agents and human review
DotOS is where I test the claims behind my writing.
What I Believe
AI should do what it is best at, and code should do the rest.
Language models are useful for ambiguity, interpretation, and refinement. Deterministic paths belong to deterministic systems.
Honesty must be structural, not aspirational.
Guardrails in prompts are weak promises. Reliable systems enforce truth through architecture, retrieval boundaries, and execution design.
The best AI systems use AI less, not more.
Good systems route AI exactly where it adds leverage and keep everything else simple, observable, and fast.
Interests
- Photography
- Classic Cars
- Biographies
- Raspberry Pi tinkering
- Architecture
- Systems thinking
Certifications / Education
- International Business Consultancy, FH Wiener Neustadt
- Professional Scrum Master
- Agile Management
- Tableau Analyst
- Certified Marketer
- PMI Certified Professional in Managing AI (PMI-CPMAI) — in progress
Contact
I’m interested in conversations about AI systems architecture, governance, and AI delivery that has to work outside the demo.