How databricks/mlops-stacks works
Databricks MLOps Stacks is a project template, initialized through the Databricks CLI, that scaffolds ML code, Databricks resources-as-code, and CI/CD workflows. The documented lifecycle is development and pull requests, automated testing in staging, deployment of merged changes to staging, then promotion to production. The tree also shows feature engineering, validation, deployment, and monitoring template areas. Source excerpts cover only a retired hook and test configuration, so relationships below rely on README statements; internal calls among generated components are not inferred.
Main entry points: resources ML job resources · workflows GitHub Actions · devops-pipelines Azure DevOps · pipelines GitLab pipelines
Components 15
ML project code
- Feature engineering
template/{{.input_root_dir}}/{{template `project_name_alphanumeric_underscore` .}}/feature_engineering - Model training
template/{{.input_root_dir}}/{{template `project_name_alphanumeric_underscore` .}}/training - Model validation
template/{{.input_root_dir}}/{{template `project_name_alphanumeric_underscore` .}}/validation - Batch inference
template/{{.input_root_dir}}/{{template `project_name_alphanumeric_underscore` .}}/deployment/batch_inference - Model deployment
template/{{.input_root_dir}}/{{template `project_name_alphanumeric_underscore` .}}/deployment/model_deployment - Model monitoring
template/{{.input_root_dir}}/{{template `project_name_alphanumeric_underscore` .}}/monitoring
Databricks resources
- ML job resources
template/{{.input_root_dir}}/{{template `project_name_alphanumeric_underscore` .}}/resources
CI/CD automation
- GitHub Actions
template/{{.input_root_dir}}/.github/workflows - Azure DevOps
template/{{.input_root_dir}}/.azure/devops-pipelines - GitLab pipelines
template/{{.input_root_dir}}/.gitlab/pipelines
Workspaces
- Staging workspace
- Production workspace
Other components
- Data scientists
- Ops engineers
- Databricks CLI
Connections 17
- Data scientists → Databricks CLI: initialize project
README.md - Ops engineers → Databricks CLI: initialize CI/CD
README.md - Databricks CLI → Feature engineering: scaffolds project
README.md - Databricks CLI → ML job resources: scaffolds resources
README.md - Databricks CLI → GitHub Actions: scaffolds workflows
README.md - Databricks CLI → Azure DevOps: scaffolds workflows
README.md - Databricks CLI → GitLab pipelines: scaffolds workflows
README.md - Data scientists → Model training: iterate on ML
README.md - Data scientists → Batch inference: iterate on ML
README.md - GitHub Actions → Staging workspace: test and deploy
README.md - Azure DevOps → Staging workspace: test and deploy
README.md - GitLab pipelines → Staging workspace: test and deploy
README.md - GitHub Actions → Production workspace: deploy changes
README.md - Azure DevOps → Production workspace: deploy changes
README.md - GitLab pipelines → Production workspace: deploy changes
README.md - ML job resources → Staging workspace: deploy ML jobs
README.md - ML job resources → Production workspace: deploy ML jobs
README.md