Code repository accompanying the O'Reilly MLOps with Databricks: Machine Learning End-to-End book by Maria Vechtomova.
This repository is organised into two self-contained projects, one per part of the book:
Covers Chapters 2-6 of the book. Demonstrates a complete ML lifecycle for a hotel booking price prediction use case, built on LightGBM, MLflow, Unity Catalog, and Databricks Asset Bundles.
| Chapter | Topic |
|---|---|
| 2 | Developing on Databricks — data preprocessing |
| 3 | Experiment tracking in MLflow, model training, logging, and registration in Unity Catalog |
| 4 | Model serving, feature serving, and endpoint authentication |
| 5 | CI/CD with Databricks Asset Bundles |
| 6 | Lakehouse monitoring |
Covers Chapter 7 of the book. Demonstrates LLMOps patterns on Databricks, including building, evaluating, and deploying an LLM-powered agent.
| Chapter | Topic |
|---|---|
| 7 | LLMOps — vector search, Genie, MLflow tracing, agent evaluation, and agent deployment |
Each project uses Python 3.12 with uv for dependency management. See the README.md inside each folder for setup instructions and environment-specific details.
# MLOps project (Chapters 2–6)
cd mlops
uv sync --extra dev
# LLMOps project (Chapter 7)
cd llmops
uv sync --extra dev