Applied AI · Industrial Systems · Brno, CZ
Building at the edge of AI and real machines.
Engineering experiments, tools, and honest notes from a CTO who has spent 20 years building software for aerospace, robotics, and industrial systems — and is figuring out what AI actually changes.
First the deterministic kind of software, now the learned kind. This site documents what actually survives contact with production.
Selected work
Process Flow Bottleneck Visualizer
The Piece That Couldn't Tell It Was Done
A generative tree whose 'am I finished?' check could return false forever, caught by a test that failed up to a third of its runs. The halting condition is part of the algorithm, not an afterthought.
The Value Function Is the Hard Part
Kaggle hands you the value function; all that's left is to optimize it. The real world never does. Whoever can define what 'better' means is already halfway to the answer — and with agents that answer for free, that gap is all that's left.
When the Model Makes the Difference
Two hours of dead ends on a routing bug — then the same prompt on the latest model fixed it in one shot. A note on not wasting time with mediocre models.