databricks-demos/dbdemos

dbdemos – Quick‑start toolkit for Databricks Lakehouse demos

What it is – dbdemos is a Python package that automates the deployment of end‑to‑end “Lakehouse” demo bundles on a Databricks workspace. The demos cover data‑engineering, analytics, and machine‑learning use‑cases (e.g., retail churn, CDC pipelines, Delta Live Tables, MLOps). The package pulls pre‑built notebooks, creates the required clusters, pipelines, dashboards and even a starter ML model, so a user can explore a full solution with a single command.

How to get it – No need to clone the repo. Install the wheel directly from PyPI inside a Databricks notebook or job:

%pip install dbdemos

Typical workflow

import dbdemos

dbdemos.help()                     # list commands
print(dbdemos.list_demos())        # see available demo names

dbdemos.install(
    'lakehouse-retail-c360',       # demo identifier
    path='./',                    # where notebooks will be written
    overwrite=True               # replace existing files
)

The call will:

1. Download the demo notebooks into the given folder.
2. Spin up a user‑specific demo cluster.
3. Run any pre‑run jobs needed to materialise sample data.
4. Create Spark Declarative Pipelines (SDP), DBSQL dashboards, and an ML model if the demo includes one.
5. Update links inside the notebooks so they point to the freshly created resources.

Key features

• One‑click demo provisioning – all required Databricks assets are created automatically.
• Support for multiple demo categories – Lakehouse end‑to‑end, product showcases (Delta Live Tables, CDC, MLOps, DBSQL dashboards, etc.).
• Customizable bundles – developers can add new AI/BI demos by providing a _resources/bundle_config notebook that describes notebooks, pipelines, workflows, and dashboards.
• Resource cleanup awareness – the tool tries to limit consumption and auto‑terminates clusters, but users remain responsible for any resources left running.
• Usage telemetry – optional reporting of demo usage (no PII) to help improve the assets.

Who needs it – Data engineers, analysts, and ML practitioners who want a ready‑made Lakehouse example to explore on their own Databricks workspace without manually wiring up clusters, pipelines, and dashboards.

Limitations / requirements

• The caller must have permissions to create clusters, SDP pipelines, DBSQL dashboards, and (for Unity Catalog demos) a UC metastore.
• Official Databricks support is not provided; issues are handled on a best‑effort basis via GitHub.

Further reading – Demo videos, the full list of bundles, and the public website are linked in the README (dbdemos.ai). For developers, see README_AIBI.md for instructions on contributing new AI/BI demos.