Trace AI Artifacts with H2O MLOps
- Observability
- H2O MLOps
- H2O Driverless AI
This lesson walks through how H2O MLOps links every registered model back to the exact Driverless AI experiment that produced it, including training configuration, comparison data, AutoDoc reports, and feature analysis. It covers how models carry scoring artifacts with them, such as the MOJO pipeline for low latency scoring and the Python scoring pipeline for more flexibility. It shows how administrators can pre-configure containerized scoring runtimes for standard, GPU enabled, or regulated environments, so data scientists deploy without infrastructure expertise. It also touches on managing experiments programmatically through an API.
Key takeaways
- How H2O MLOps links a registered model back to its exact Driverless AI experiment and training configuration
- What artifacts travel with a model, including AutoDoc reports, MOJO pipelines, and Python scoring pipelines
- Why pre-configured containerized runtimes let data scientists deploy without infrastructure expertise
- How system administrators set up separate runtimes for standard, GPU enabled, and regulated workloads
- The API lets teams query and manage experiments linked to registered models programmatically
Start owning your AI