Skip to content

Project Spotlight

Pittsburgh Paints

ML Paint Matching

Introduce AI and prediction to improve MES processes

    • Built API functions for processing machine-learning vectors for manufacturing in C# (.NET Core, ML.NET, EF Core).
    • Ingested data for the new predictive-analytics pipeline via Azure storage ( Dataverse KQL and Azure SQL ).
    • Expanded the database and stored procedures in T-SQL (Azure SQL).
    • Loaded the existing color and additive dataset into Azure Data Lake and Microsoft Fabric for processing.
    • Use Siemiens Opcenter MES API to provide mixing data and update the Data Lake.
    • Built Python clients to consume the AI REST web services.
    • Trained a forward-regression model with Azure AutoML, applying guardrails to prevent waste and maintain safety margins.
    • Contributed to product roadmaps for future development and deployment to Azure App Service via Azure DevOps.
    • Built C# ASP.NET Core REST endpoints that accept current mix/color data and return a corrective formula to reach the target color.
    • Built services to periodically retrain the model so it continues to mirror real-world scenarios.
    • Monitored services and produced performance reports in Azure Monitor using KQL to query metrics
      •