Services
I build software that turns data into decisions, whether for a bakery, a workshop or a public administration.
My craft comes from critical infrastructure: in my main role I develop the control system of a hydropower plant. There, a prototype in a notebook is not enough. I bring that standard to every project.
Typical questions I solve #
How much should I order tomorrow?
Daily demand forecasts per product for a regional bakery. The model uses the bakery’s own sales data, plus weather, school holidays and weekdays. In a three-month backtest the return rate dropped from 15-25 percent to 10-15 percent, without shelves running empty at lunchtime more often; the system is currently in supervised live testing.
What do aerial images and geodata tell us?
Analysis of aerial imagery and survey data, for example for municipalities, planning offices or utilities. Detecting surfaces, counting stock, tracking change over time.
What is hiding in our numbers, and how do we see it every day?
Data preparation, analysis and dashboards built around a concrete question. Often the fastest way to start: a few days of effort, immediately useful in daily operations.
Example project: an interactive comparison tool across 2,000 Swiss municipalities →
And in brief:
Detect cracks, knotholes or dimensional deviations right on the line.
Spot anomalies in sensor and log data while there is still time for planned maintenance.
Extract invoices, receipts and forms automatically, with uncertain cases reviewed by a human.
Search contracts, meeting minutes and manuals in plain language, with source references.
Who is likely to churn and where extra attention pays off.
Optimization that tries millions of variants before the dispatcher sees the plan.
Your case is not listed? It is usually solvable anyway. A first conversation costs nothing →
How I work #
First check whether it is worth it
At the start we clarify which data exists, which decision should improve and whether the effort will realistically pay off. If I think the project is a bad idea, I will tell you.
One point of contact
From data preparation to the model to the system running in production. You coordinate no team and no agency.
Built for low maintenance
I prefer a simple model that runs stably for three years over a complex one nobody wants to touch anymore. Documented and handed over cleanly, so you do not depend on me.
Your data stays your data
Processing in your environment or in Switzerland wherever possible. Confidentiality goes without saying, with an NDA on request.
Process and scope #
First conversation free, around 30 minutes
Problem, data situation, rough feasibility.
Assessment free, as long as it stays manageable
Review of the data, a concrete solution proposal, effort estimate.
Experiments
Short iterations on your data. After each one we discuss the results and decide together how to continue. If something does not work, you see it early, not on the final invoice.
Deployment
The best approach becomes a system that runs in daily operations.
Handover
Documented, with an operations manual. Maintenance on request.
Typical project sizes range from a few days (analysis, automation) to several weeks (a model in productive use). Hourly rate or fixed price by agreement. I work remotely; my base is the canton of Schwyz. Freelancing runs alongside my employment and my studies; I therefore deliberately take on only a few projects at a time.