Who I am,
and how I work.

Making music is learning the rules, and knowing when to depart from them.

Composing and producing has two sides. The musical side is a set of rules for harmony, rhythm and structure; you learn them so you can deliberately let them go. The technical side is the signal path from instrument to speaker, in which every link has to pass on exactly the right level.

Besides that, I think about the listener with every track. When preparing a DJ set that is the whole job: who is in the room, what mood are they in, and which next record takes them where I want them to go.

Making images by operating a calculation by hand.

I make abstract images with a calculation called convolution: the same operation photo software uses to sharpen or blur an image. I operate it by hand, through a converted cash-register keypad, and repeat it until an image appears that keeps my attention.

The operation itself is nothing special. It runs in every phone camera and in every model that recognises images. The difference is that I steer it myself instead of letting it run automatically.

Taking a good photo with as little equipment as possible.

For me, photography is capturing a moment with as few means as possible.

At events I shoot with a Canon from 2013 and a wide-angle lens, and the results hold up next to those from far more expensive cameras. Product photos for web shops I take with a Sony Mavica from 1998. It produces small files of 640 by 480 pixels that need no processing at all on a website and still look modern.

After the shoot comes the selecting: from hundreds of photos down to a series of ten.

Printing digital images without losing anything.

I print my images myself on modified inkjet printers. That came from a requirement: the images are small, from 64 by 64 to 512 by 512 pixels, and on paper every pixel has to stay visible as a sharp square.

A normal printer softens edges when enlarging. For this work that has to be switched off. Then comes colour management: making sure the colour on paper matches the colour on screen. To me that is a necessity, not a hobby, but I do get the most out of it with cheap ink and cheap paper.

I measure colours with a professional colorimeter I bought as defective. With open-source software it works again at the level of a new unit.

Electronics in which you can see and measure every part.

Analog audio equipment is fun because you can see and measure everything. Every part has one function, and with a meter you find where the fault is without a manual.

Often there is mechanics in there that solves what software does today. An old cassette deck reverses the tape automatically with a system of levers and gears; a newer deck does the same with a second head and a bit of code.

Programming machines that produce for days on end without supervision.

As an apprentice machinist I programmed and operated CNC milling machines and the robots that feed them. The real work is the program: it has to keep the machine running for days with as little downtime as possible.

So you plan for what goes wrong. A cutter wears, and the program has to correct for that before the dimensions drift. You measure along the way and adjust. And every second counts: one second per piece is almost three hours on a run of ten thousand.

A 3D printer is a tool, not a hobby.

I design in CAD and 3D print because it is useful, not because I love it.

It is the fastest way to replace a part or hold an idea in your hands: drawn this afternoon, printed tonight. As with the inkjet printer, I get more out of cheap equipment here than the manufacturer promises.

Flying a drone that cannot stop.

I fly racing drones in manual mode. In that mode the drone does not stop when you release the throttle: it keeps going at its speed, and for a simple turn you operate four axes at once.

So you have to look ahead and correct continuously. That takes full concentration, and that concentration is exactly why I do it.

First understand why something is broken, only then repair it.

With my own things I repair by feel: open it, look, fix it. That almost always works. I follow a fixed procedure when the stakes are high, as with the Apple repairs I did at Coolblue, or when the cause is not immediately clear.

What I enjoy is finding the cause: why does this device do this. Replacing a screen or battery is routine. Understanding why a device keeps rebooting is the real repair.

How a brain works, and what that says about AI models.

What occupies me is the similarity between a brain and a language model. Your thoughts arrive on their own and you only partly steer them, much like the text a model produces word by word. And a mood works like an instruction given in advance: whoever starts out gloomy reads everything that follows as gloomy. A model does the same with the context you give it.

I read about it out of interest, but I use it daily.

Explaining in one sentence what would otherwise take an hour.

I like to explain things with a comparison. The yardstick: my mother has to understand it. The aim is to reduce something complicated to the shortest explanation that is still correct.

I learned that at the service desk at MediaMarkt and Coolblue, where I had to explain to customers why their problem was not covered by the warranty. That only works if you truly understand it yourself.

Looking at something far bigger than your own work.

Astronomy gives me perspective. Everything that exists, from stars to people, comes out of the same natural processes. That is the largest system there is, and I find it calming to look at it now and then.

Spaceflight is the technical side of it. The first probes were controlled from millions of kilometres away with less computing power than a Game Boy. That it worked came down to careful design, not expensive equipment.

This is where all the interests come together.

Since mid-2025 I have been building websites, shops and automations full-time, in TypeScript and Python, together with AI models such as Claude. Everything I learned in music, metal, image and language goes into that.

The work consists of the same steps: working out how something works, designing it, building it, checking it critically, repairing it until it is right, and explaining what is happening.