Infrastructure that documents itself
The systems everything else runs on — designed as code, backed up, and recoverable. Built to be understood a year later, not just today.
§Infrastructure & AI Systems Architect
Platforms rigorous enough to run unattended.
I design, build, and operate infrastructure and AI systems that are made
to be trusted — self-hosted, dependable, and reasoned about in the open. The quiet kind of
engineering: built with the discipline of things that can't afford to fall over.
- Role
- Infrastructure & AI systems architect
- Focus
- Platforms · reliability · AI & LLM systems
- Based
- Prague, Czechia · working across CS / EN
- Contact
- martin@stanek.fun
01/ What I do
Four things, done properly.
I work across the whole stack — from the machines up to the model — and care most about the parts that decide whether a system can be trusted.
Platforms
Reliability
Systems that watch themselves
Making infrastructure legible and self-observing, so problems surface — and get caught — before anyone has to notice them.
AI systems
AI platforms grounded in evidence
Building and evaluating AI and LLM systems where the choices — what to run, how, at what cost — rest on measurement rather than hype.
Security
Networks with the posture written down
Secure, well-segmented networks whose rules live in one place and are continuously checked, not assumed.
02 / Approach
How I build.
Principle 01
Decide in the open.
Every non-trivial choice is written down — options, tradeoffs, and the reason on file. Anyone can see why, not just what.
Principle 02
Verify, don't assert.
Nothing is “done” until something independent proves it — a test, a probe, a real round-trip. A clean log is not a working system.
Principle 03
Keep the data close.
Self-hosted by default; the cloud is a deliberate, weighed exception, not a reflex. Own the platform you depend on.