I'm Jonas. THE Software Engineer.

Then the wood. My first automation, and I no longer have to chop down trees. More time. So the next automation. And the one after that. At some point the robot simply does everything. And now? What do I do with all that time? Plugins. Minecraft plugins, in Java. My first time with object-oriented programming, and honestly: I understood almost nothing. Until, on a school trip (a weekend away with my class), I read my first book on Java and the basics of programming. Suddenly: Ah. Data types. That actually makes sense.

AI as we know it today didn't exist back then. Instead there were the guys from India in the YouTube videos, with an English accent and call-center audio quality. Vibes. The best part: they still help me to this day.

Then vocational college as an IT assistant, and after that the apprenticeship as an IT specialist. I'm sitting in vocational school, and for the umpteenth time here comes the example: object orientation and the blueprint of a house. The class is the blueprint, the object is the house. One blueprint, any number of houses. With windows, doors, a roof, maybe a chimney with smoke on top. Please, not the house again. My robot in Minecraft would have built it long ago, block by block, according to plan. I'd just have to press Enter.

Then, during the apprenticeship, my first really big project: ERP software with thousands of features, several developers, real customers. Built in C# with the .NET Framework and WinForms, plus a relational database on Microsoft SQL Server. We organized ourselves with Scrum. And I was right in the middle of it. At first small changes and scripts. Understanding how everything fits together. Then bigger adjustments, more responsibility and the first interfaces. Sometimes to a high-bay warehouse, sometimes to other systems. Sometimes just a file on an FTP server. Not everything has to be spectacular. It has to work and make the customer's job easier.

Alongside that, my side job, which I started during the apprenticeship. There I worked on an older repricing software for the B2B sector, written in Python and Perl. Adjusting prices automatically instead of by hand. My Lua robot in Minecraft took the tree chopping off my hands. This software took manual repricing off the customers' hands. Different task, same idea. Suddenly Python was no longer a snake but a programming language. And the camel? Perl.

Fully trained. At least on paper. In reality, you never stop learning in this job. As soon as you've understood something, the next question is already knocking. That's exactly what I like about it. Vocational school was over, and instead the really big changes were waiting for me in the development team. More responsibility, more connections I had to understand. The house blueprint had finally served its purpose. Learning hadn't.

And today? More than ten projects in four years. My current toolkit: TypeScript with Node.js, SQL, n8n and Python. Depending on what the task needs. A lot of it revolves around automation: exchanging data between systems, connecting processes, taking manual work off people's hands. Basically, my robot from back then is still at it. Only today it doesn't chop down trees anymore, it saves people the same old manual steps.

Backend engineer, data engineer, automation engineer. Pick one, Jonas. Not that easy when you'd rather help build everything. But I needed a focus. Today mine is backend and automation. So: backend & automation engineer.

On the iceberg, that would be the part below the water. Processing data, connecting systems, keeping processes running. Everything that has to work in the background so someone up top can press a button and the right thing happens. Sometimes I surface and help out on the frontend. More in the past, less these days. Still, I like knowing what happens above the waterline.

My focus is less on UI and UX and more on DX: developer experience. In other words, how well other developers can work with what I build. Is the interface understandable? Does the error message help? Or is someone sitting there thinking: What on earth did Jonas do here?

I like it when things just work. When unit and integration tests cover "all" test cases. At least the ones I thought of. When the frontend can rely on the backend. And when changes are automatically tested and rolled out through a CI/CD pipeline, instead of going through the same steps by hand for every deployment.

And AI? It was bound to come at some point. Working with it, though, was something I had to learn first. In the beginning it was more like an overeager minion. Highly motivated, fast-moving, and I constantly had to keep an eye on it. What are you doing? No. Not that file. Please ask first.

Over time I learned to give it not just tasks, but the context that goes with them. README files that explain the project. Skills and commands for recurring workflows. Agents with clearly defined tasks. And above all: not everything at once. Small steps, test, review, refine. Pretty agile, actually.

Today AI can already take over a lot with Claude Code and workflows like these. I set the direction, review the results and decide what stays. With n8n, Python and LLMs I also automate whatever keeps repeating and costs time. Mine and other people's.

The same trick as back then with the wood. If you no longer have to chop down every tree yourself, you have time for the next build. For me that means: what really needs focus gets focus. The rest runs on its own. And when everything runs? Then I stay on it. And keep it running.