energy-forecasting

Why we included this project

For engineers tired of isolated MLOps tutorials, this repo ships a full batch system you can actually run. The seven-lesson course trains an hourly energy-demand model for Denmark and carries it through the machinery of production ML, from a feature store, experiment tracking, and Airflow-orchestrated training and prediction to monitoring and serving through FastAPI and Streamlit, finishing with a GCP deployment. Because the working code comes with about two and a half hours of written and video lessons, it works both as a transferable blueprint for your own batch pipelines and as a hands-on learning path for ML engineers and developers moving into MLE. The energy domain is the concrete example, but the habits worth copying, data validation, versioning, Poetry packaging, and CI/CD, are what stay useful after you put the repo down.