Applications

Data science, AI & automation

My actual field. I solve problems with code ^^ from data analysis to custom software.

Schematic: long report, short summary On the left a stack of report pages full of text lines; an arrow labelled AI leads to a small card on the right with five highlighted key points.
fig. 1 — schematic: 40-page report, 5 lines of substance

Focus areas

Data Science & Analytics

Turn your data into insight. Analyses, forecasts and dashboards, prepared as a basis for decisions, so you can decide better.

Python Pandas Demand Forecasting Decision Intelligence

Machine Learning

Models that find patterns in your data, for example which customers are unlikely to come back or which products are often bought together.

Scikit-learn TensorFlow

AI Solutions

Practical AI applications, such as automatically condensing long reports down to what matters. But also solving complex tasks.

LLMs Prompting Prompt Caching

Automation & Scripts

I handle recurring tasks automatically with Python scripts: merging data, generating reports, connecting systems.

Python Scripts

Custom Tools

Tailor-made small programs instead of unwieldy Excel sheets: quote generators, internal tools, administration systems.

Web apps Custom software

Interfaces & APIs

I connect your existing programs and services so that data flows automatically to where you need it.

APIs Integration
Demand forecasting · try it yourself

How many bread rolls on Saturday?

A made-up bakery, a year of made-up sales figures – and a real model that was trained in your browser just now. No server, no customer data. Pick the day:

Day of the week
Weather

– bread rolls

Sun Clouds Rain Public holiday forecast with 80 % interval

The data is made up, the model is real – see demo.js. How it works is explained in the notebook.

Projects

NetSales model

Analysis and forecasting model for sales data: data cleaning, key figures and predictions that make sales trends visible early.

PythonPandasForecasts

ML projects

Classification, clustering and recommendation systems on structured data, from exploration to production model.

Scikit-learnTensorFlowMLOps

Testing suite for websites

Automated test pipeline for web projects: E2E tests, visual regression, performance budgets, runs on every deploy.

PlaywrightCI/CDQA

AIOps

Automated operations: anomaly detection in logs and metrics, incident routing, self-healing scripts for cloud infrastructure.

PythonMonitoringAutomation

Bread roll forecast

Demand forecasting to try out: a model that trains right in your browser, on made-up data for a made-up bakery. It shows the forecast with an 80 % interval and how far off it lands on test weeks it has never seen. In the notebook I explain how it works.

JavaScriptDemand ForecastingPrediction interval

This website

Its own design instead of a template, no cookie banner. On every change, automatic tests check the HTML validation, all internal links, how the pages render in a real browser (phone and desktop, with and without JavaScript) and the Lighthouse scores. How it all fits together is on How it’s built.

PlaywrightGitHub ActionsLighthouse

Neural network in Blender

An 8-second loop rendered in Blender: a forward pass through 71 neurons and 266 connections. It runs on the home page, only loads once it is in view, and stays a still image with reduced motion.

BlenderNeural networks

A look at my work: GitHub →

Next step

Is a process eating your time every week?

Tell me which work keeps repeating itself. If it can be automated, I will tell you how — and if it cannot, I will tell you that too.

Send an enquiry Write an email