Project Information
- Category: Municipal Infrastructure & AI
- Role: Project Manager (Prosjektleder)
- Client: Bergen Vann
- Stack: Azure Data Factory, Azure Databricks, Power BI
- Links: Bergen Kommune article, Bouvet article
Machine Learning for Smarter Wastewater Management
This project aims to optimize Bergen Vann's wastewater infrastructure using machine learning and cloud-based data pipelines.
The system collects real-time data from millions of sensor readings via Azure Data Factory, primarily from the city’s 260+ pumping stations. Data is processed and prepared through automated ETL pipelines, enabling efficient feature engineering and model training in Azure Databricks.
Outputs from the models are integrated into Power BI dashboards that provide insight into overflow risks, pump behavior, and deviations from expected flow – making it easier for operations staff and engineers to monitor and act on relevant changes.
My role has focused on improving data quality, restructuring the pipeline logic, and updating the ML models to increase prediction accuracy and robustness. I’ve also implemented smarter alerting and system documentation to support long-term maintainability.
Additionally, I've developed and maintain over 15 Power BI reports across the organization, supporting everything from leakage detection to climate reporting – often by combining internal data streams with external sources such as Fjordkraft.