Dynamo Examples, Tutorials and Deployment guides

🚧 Work in Progress

This repository is still a work in progress. More comprehensive examples and documentation can be found at the main examples repository: https://github.com/ai-dynamo/dynamo/tree/main/examples

About Dynamo

Dynamo is an AI inference platform designed for high-performance, scalable deployment of machine learning models. These examples demonstrate various deployment patterns, from simple single-node setups to complex multi-stage pipelines and distributed architectures.

Note

⚠️ This project will download and install additional third-party open source software projects. Review the license terms of these open source projects before use.

Directory Structure

basics/

Foundational examples to get started with Dynamo:

• hello_world/ - Simple hello world example demonstrating basic Dynamo component structure and execution flow
• hello_world_configurable/ - Configurable hello world example showing how to parameterize components and use configuration files
• hello_world_multinode/ - Multi-node hello world example demonstrating distributed execution across multiple nodes with:
  • Component definitions in components/
  • Configuration files in configs/
  • Visual diagrams in _img/
• multistage_pipeline/ - Multi-stage pipeline example showcasing complex data processing workflows with:
  • Modular components in components/
  • Environment-specific configs in configs/
  • Pipeline graphs in graphs/
• simple_pipeline/ - Streamlined pipeline example for understanding core pipeline concepts

llm/

Large Language Model examples and configurations:

• SNSvLLM_Disagg_SingleNode/ - SNS vLLM disaggregated single node example with visual documentation in images/

customizations/

Advanced examples showing custom implementation patterns:

• Custom component development
• Extending Dynamo functionality
• Integration patterns with existing systems

deployments/

Production deployment examples and configurations:

• Cloud deployment strategies
• Container orchestration setups
• Scaling and monitoring configurations

Getting Started

1. Start with Basics: Begin with the hello_world example to understand core concepts
2. Explore Configurations: Try the hello_world_configurable to learn about parameterization
3. Scale Up: Move to HelloWorld_MultiNodes for distributed scenarios
4. Build Pipelines: Experiment with simple_pipeline and multistage_pipeline examples

Additional Resources

• Main Examples Repository: https://github.com/ai-dynamo/dynamo/tree/main/examples - Comprehensive collection of production-ready examples
• Documentation: https://github.com/ai-dynamo/dynamo/tree/main/docs - Full documentation and guides
• SDK Documentation: https://github.com/ai-dynamo/dynamo/tree/main/deploy/sdk/docs - SDK reference and CLI guides

Support

For questions, issues, or contributions, please visit the main Dynamo repository and check the existing issues or create a new one.