0. Image-wise Classification
1. GPS Analytics
2. Spark and Jupyter Notebook
3. Kubeflow on Kind cluster
4. Example kubeflow for Sentinel 2 Image-wise Classification
5. License

• Starter notebook for image-wise classification tasks
• Feature engineering techniques to improve baseline models
• Different flavours of global interpretability
• In-Depth Model Analysis and Debugging (inprogess)
• End-to-end ML pipeline using Kubeflow for scalable workflows

• Starter notebook for GPS processing with Polygons broadcasting
• Streamline timeseries forecasting with Feast

You can find a detailed step-by-step guide in:

notebooks/gps-analytics/README.md

Otherwise, for one-shot deployment (from within notebooks/gps-analytics):

./deploy_docker_spark_jupyter.sh "$PWD"

accessing PySpark inside the container:

docker exec -i -t <name-container> /usr/local/spark/bin/pyspark

access Spark UI:

http://localhost:4040

You can find a detailed step-by-step guide in:

notebooks/image-classification/README.md

Otherwise, for one-shot deployment (from within notebooks/image-classification):

./deploy_kubeflow_pipeline.sh

Check pods status:

./check_pods_status.sh

Shutdown the cluster:

./shutdown_kind.sh

• Ensure Docker and Kind are installed and running
• Deployment may take a few minutes depending on your system
• Use the pod status script to wait until all services are ready

Check End-to-end ML pipeline using Kubeflow for scalable workflows