DataTwinLabs● Selected work · 08
Built,shipped, used.
Every project here started with a real problem and ended in something people use. Problem, build and result, for each one.
01 · Flagship · KIVI Award finalist
City-scale digital twin of 's-Hertogenbosch
A streaming lakehouse that fuses building registry, traffic, weather and air sensors into one live 3D model of the city.
All projects
The experience behind the work
Three disciplines, one engineer.
Senior data engineer, building production pipelines today.
Batch and streaming ingestion, dbt models, warehouses and lakehouses on GCP and Azure. Tested, monitored, documented.
10 years of building software, including leading a team of five.
Java enterprise, Grails/Groovy, REST and service-oriented design, microservices, design patterns, dependency injection. Fintech, banking alerts and e-commerce systems.
Models that reach production and stay healthy there.
Forecasting, computer vision and retrieval assistants. Experiments tracked, models registered, deployed behind APIs, monitored and retrained through CI/CD.
What clients come to me with
Your problem, the tools that solve it.
“Our data is everywhere and nobody trusts the numbers.”
One governed platform: ingestion, a tested warehouse or lakehouse, and a single source of truth.
“We have a model in a notebook. It never reaches production.”
Models tracked, versioned and deployed behind an API, with monitoring and retraining.
“We want to ask our documents and data questions.”
Retrieval-augmented assistants grounded in your own sources, with guardrails on what they may say.
“We need live sensor or traffic data on a map, not a monthly report.”
Streaming pipelines into digital twins and dashboards people open every day.
“Our software is hard to change and deploys are scary.”
Clean services, automated tests and pipelines, infrastructure as code.
How it fits together
From raw data to models in production, on one platform.
A typical end-to-end setup I design and build on Azure: ingestion, a Databricks lakehouse, machine learning tracked in MLflow, dashboards, and the DevOps and MLOps that keep it all running. The same shape works on Google Cloud or AWS.
What you keep
Things you own when we're done.
Architecture diagram + decision log
Pipelines in Git, with tests and alerts
dbt models and a documented data model
Trained models in an MLflow registry
APIs with OpenAPI docs
Dashboards your team actually uses
Terraform / CI-CD so it rebuilds itself
A handover session and runbook
● Next project