This hands-on workshop demonstrates how to build and deploy secure data and AI applications directly on the Databricks Data Intelligence Platform. Databricks Apps provides serverless hosting with deep integration across Unity Catalog, Databricks SQL, Model Serving, and Lakeflow Jobs—eliminating separate infrastructure while providing built-in governance, identity, and observability.
Key Benefits:
- Simple Development: Support for Python frameworks (Streamlit, Dash, Gradio) and Node.js; develop locally or in-workspace
- Built-in Security: Unity Catalog permissions, managed service principals, OAuth 2.0 SSO, and optional on-behalf-of (OBO) user authorization
- Production-Ready: Serverless compute with unique URLs, Git/CI-CD support, and no infrastructure management
Use Cases Unlocked:
- Intuitive data explorations with governed access controls
- Secure GenAI chat applications with hosted models
- Operational forms and analytics interfaces
- Model endpoint exposure through secure UIs with consistent governance
Participants will complete three progressive demos showcasing different Databricks Apps capabilities:
Framework: Streamlit
Duration: 15-20 minutes
Focus: Model Serving integration and basic app deployment
Build a PDF text extraction application that:
- Uploads and processes PDF documents using Vision AI
- Integrates with Databricks Model Serving endpoints
- Demonstrates automatic OAuth authentication
- Exports results to CSV, text, or Delta tables
Key Concepts: Streamlit development, Model Serving endpoints, SQL Warehouse connectivity, environment variable injection
Framework: Dash (Plotly)
Duration: 20-25 minutes
Focus: Transactional workloads with Lakebase and Row-Level Security
Build a support ticket system that:
- Submits and manages tickets with status tracking
- Displays tickets in a kanban board interface
- Enforces PostgreSQL Row-Level Security (RLS) at the database level
- Uses on-behalf-of (OBO) authorization to propagate user identity
Key Concepts: Databricks Lakebase (PostgreSQL), Row-Level Security, OBO authorization, CRUD operations
Framework: Dash (Plotly)
Duration: 25-30 minutes
Focus: Direct Databricks SQL integration and AI-assisted development
Build a note-taking application that:
- Connects directly to Databricks SQL using the SQL Connector
- Creates and manages Delta tables with identity columns and timestamps
- Deploys via Databricks Asset Bundles (DABs)
- Uses Model Context Protocol (MCP) for AI-assisted schema modifications
Key Concepts: Databricks SQL Connector, Unity Catalog three-level namespace, Delta Lake features, Infrastructure-as-Code with DABs, MCP for AI-assisted development
- Access to a Databricks workspace with Apps enabled
- Databricks CLI installed and configured
- For Demo 1: A Model Serving endpoint with a vision-capable model
- For Demo 2: On-Behalf-Of User Authorization enabled (workspace admin setting)
- For Demo 3: A SQL Warehouse and Unity Catalog access
- (Optional) AI tool with MCP support for Demo 3 advanced features
databricks-apps-workshop/
├── README.md # This file
├── databricks.yml # Root Asset Bundle configuration
├── deploy_app_bundle.sh # Deployment helper script
├── apps/
│ ├── demo1_pdf_extractor_streamlit/
│ │ ├── app.py # Streamlit application
│ │ ├── app.yml # App-specific configuration
│ │ ├── requirements.txt
│ │ └── WORKSHOP_USER_GUIDE.md # Step-by-step instructions
│ ├── demo2_support_tickets_dash/
│ │ ├── app.py # Dash application
│ │ ├── app.yml
│ │ ├── setup-lakebase.ipynb # Database setup notebook
│ │ ├── requirements.txt
│ │ └── WORKSHOP_USER_GUIDE.md
│ └── demo3_dash_dbsql/
│ ├── app.py # Dash application with SQL Connector
│ ├── app.yml
│ ├── requirements.txt
│ ├── WORKSHOP_USER_GUIDE.md
└── dab/
└── resources/
└── apps.yml # Asset Bundle app definitions
git clone https://github.com/dbderek/databricks-apps-workshop.git
cd databricks-apps-workshopInstall databricks CLI with Homebrew
brew install databricksConfigure the CLI
databricks configure --profile workshopFollow prompts to enter your workspace URL and authentication.
Each demo is self-contained with its own WORKSHOP_USER_GUIDE.md. Navigate to the demo folder and follow the guide:
- Demo 1:
apps/demo1_pdf_extractor_streamlit/WORKSHOP_USER_GUIDE.md - Demo 2:
apps/demo2_support_tickets_dash/WORKSHOP_USER_GUIDE.md - Demo 3:
apps/demo3_dash_dbsql/WORKSHOP_USER_GUIDE.md
Before deploying, update the configuration files with your workspace details:
- Root
databricks.yml: Workspace host URL - App-specific
app.yml: Resource IDs (warehouse, endpoints, etc.)
Each demo uses Databricks Asset Bundles for deployment:
databricks bundle validate -t dev
./deploy_app_bundle.sh devRecommended Order: Complete demos sequentially (1 → 2 → 3)
- Demo 1 introduces core concepts: app deployment, authentication, and Model Serving integration
- Demo 2 adds complexity with transactional databases, RLS, and OBO authorization
- Demo 3 demonstrates advanced patterns: direct SQL integration, Infrastructure-as-Code, and AI-assisted development
Total Time: 60-75 minutes for all three demos
- Databricks Apps Documentation
- Databricks Asset Bundles Guide
- Unity Catalog Overview
- Model Context Protocol (MCP)
For questions or issues during the workshop:
- Check the
WORKSHOP_USER_GUIDE.mdtroubleshooting sections - Review the Databricks Apps documentation
- Contact your workshop facilitator
This workshop content is provided for educational purposes.