Work
I've spent fifteen years between engineering and the people who have to decide things. The technical half is data platforms, pipelines and AI systems. The other half is sitting with the people who own the problem until the requirement is actually a requirement and not a wish.
Currently placed at Tetra Pak through Accelerated Growth AB.
Tetra Pak
Dairy production data
A dairy operation generates data at every stage: collection, separation, pasteurisation, whey streams, cheese, filling and packing. Each stage has its own systems and its own people. Very few can see across all of them at once.
I work across that chain as a data engineer and developer. Pipelines that bring the stages into one place, dashboards that operations teams actually open, and machine learning models and AI reporting built on top of the result.
- Built with
- Azure Databricks, Unity Catalog, Lakeflow, Databricks SQL, MLflow, Power BI.
Tetra Pak
Sustainability recommendations
Two problems landing on one department. The sustainability specialists knew there were improvements available across customers' production lines that nobody was finding systematically. They were also answering every specialist question coming from a sales organisation working across more than 160 countries.
I built a recommendation engine that identifies sustainability options for a given customer, covering both changes to the production line and broader measures like solar. And a chatbot trained on internal sustainability material, so the global team can get answers on specialist subjects without routing every question through a department that was never going to scale to meet them.
- Built with
- Azure Databricks, Power BI, Dash, Azure AI Foundry with RAG over internal sustainability material.
Accelerated Growth AB
Talent matching
Our commercial inbox was the bottleneck. Every assignment request arriving needed evaluating against what we could actually staff, and once it qualified, someone had to work through the consultant database weighing availability, location, skills and where that person wanted their career to go. It isn't work that gets faster by adding people to it.
I built a system that reads the inbox, qualifies incoming requests and ranks consultants against them on those criteria. It cut a process measured in hours down to minutes.
This is the one I can show in detail, because it's ours rather than a client's.
- Built with
- Azure Logic Apps, deployed models.
IKEA
Engineering management
Engineering management and vendor strategy for the platforms behind IKEA's organisation-wide surveys. Line management of a software engineering team, four suppliers across different products and services, budget ownership, and the integration work to make several separately-bought solutions behave like one thing.
Less AI than the rest of this page, and more of what makes AI projects actually land: getting engineers, suppliers and product management pointed at the same outcome before anyone writes code.