- simple train, eval, deploy with SDK
- orchestrate with Vertex pipelines
- compare multiple forecast scenarios
- compare multiple forecast modeling approaches (e.g., BQML Regression, ARIMA+, Vertex Forecast, FB Prophet, etc.)
- Model Versioning AutoML models with Vertex AI Model Registry (see example here
- challenger vs blessed methodlogy for production deployments (see example here
- forecasts that include external and/or public data (e.g., GDELT)