Software developer · Germany
I build web products, mostly with Laravel. Before I write the code, I like to work out the flow, the data and the architecture behind it.
Professional work
A Laravel backend that received complete datasets from an internal business system and turned them into structured application data. Incoming codes and inconsistent values had to be matched, cleaned and mapped into proper relationships before the rest of the product could use them.
On the outgoing side, the same backend served the processed data through an API and handled caching, scheduled exports and recurring reports.
A Laravel admin application paired with an Ionic/Vue mobile app. Different user roles worked through registrations, approvals, statuses and notifications across the two interfaces.
I worked mainly on the mobile app, building the workflows volunteers used to browse activities, register and move through the approval process.
Across agency projects, the same work kept coming up: APIs, auth and permissions, admin tools, queues, imports, integrations, reporting and notifications.
I usually had a lot of freedom on the technical side. I often chose the libraries, the API structure and the architecture myself, and decided how background jobs and admin tooling should work.
How I work
What is the person using this actually trying to get done, and what’s in their way?
The flow, the states, the data and the technical boundaries. What runs in a request and what goes on a queue.
The part I enjoy most. The plan usually changes a little once real data shows up, and that’s fine.
Laravel is where I have the most experience and where I move fastest. But I don’t particularly care whether every problem ends up being solved with PHP. If one leads into Linux, networking or a tool that’s better written in Go or Rust, I’m happy to follow it there.
Currently
AI is changing both the software we build and the way we build it. Most of what I’m working on sits somewhere in that.
A mobile app and Laravel API for tracking grocery prices. OCR runs in a queued job, uncertain matches go back to the user, and every line becomes a price for a product, store and date. The OCR is one step. The product is the price history.
Agents are a big part of how I work now. I’m trying to work out which tasks they can really take over, how my workflow changes around them, and what kind of environment they need, which has led me into remote Linux workspaces.
A small self-hosted server provisioning tool. Mostly an excuse to understand what happens below the deploy button: systemd, Caddy, databases and deployment lifecycles.
I’m looking for a team where developers get to own a problem, not just the ticket for it. Small or fairly autonomous product teams tend to suit me best, and I care a lot about working with kind people.
I’d also like AI to be a real part of the work, either in the product or in how we build it.