Ten years ago I was writing LabVIEW code on an FPGA to measure blade vibrations in a 390 MW turbine. Today I’m orchestrating AI agents in Kubernetes to triage 5G Core test failures.

The tools are completely different. The fundamentals are the same.

What Actually Changes

The hardware disappears. In industrial automation, you’re fighting latency, hardware reliability, and the physics of sensors. In cloud-native work, the infrastructure is abstracted — you fight distributed systems complexity instead.

The feedback loop speeds up. Deploying a fix to a physical test machine on a factory floor might take a day. Deploying to cloud takes minutes. This changes how you think about iteration.

The data volumes explode. A turbine measurement system produces gigabytes. An IIoT platform collecting from hundreds of robots produces terabytes. The same data engineering principles apply — but the operational complexity is orders of magnitude higher.

What Stays the Same

Reliability is non-negotiable. A turbine diagnostic system that crashes during a critical measurement is useless. A cloud platform that loses data during edge collection is useless. The tolerance for unreliability is zero in both worlds.

Users don’t care about your stack. Factory engineers care whether the vibration plot updates in real time. Robot operators care whether the diagnostics portal shows accurate data. The technology is a means to an end.

You need to understand the domain. You can’t build good industrial software without understanding what a tip-timing measurement is. You can’t build good 5G software without reading 3GPP specs. Domain knowledge is a force multiplier.

The Bridge: IIoT

Industrial IoT sits exactly at the intersection. At KOGENA we were extracting data from robots using OPC-UA, CAN bus, and proprietary protocols, pre-processing it on edge hardware, and streaming it to a Kubernetes cluster in the cloud.

You need to understand both worlds simultaneously — or your system falls over at the seam between them.

A decade in, I’m still most interested in the hard problems at the boundary: where AI meets systems engineering, where cloud meets the physical world.

Wrocław, 2024