Databricks Core Workflow B: MLflow Training

Overview

Build ML pipelines with MLflow experiment tracking, model registry, and deployment.

Prerequisites

• Completed databricks-install-auth setup
• Familiarity with databricks-core-workflow-a (data pipelines)
• MLflow and scikit-learn installed
• Unity Catalog for model registry (recommended)

Instructions

Step 1: Feature Engineering with Feature Store

Step 2: MLflow Experiment Tracking

Step 3: Model Registry and Versioning

Step 4: Model Serving and Inference

For full implementation details and code examples, load: references/implementation-guide.md

Output

• Feature table in Unity Catalog
• MLflow experiment with tracked runs
• Registered model with versions
• Model serving endpoint

Error Handling

Resources

• MLflow on Databricks
• Feature Engineering
• Model Serving
• Unity Catalog ML

Next Steps

For common errors, see databricks-common-errors.

Examples

Basic usage: Apply databricks core workflow b to a standard project setup with default configuration options.

Advanced scenario: Customize databricks core workflow b for production environments with multiple constraints and team-specific requirements.