The Steam AI engine is an end-to-end platform that streamlines the entire process of building and deploying smart applications. Now data scientists and developers can launch turnkey compute environments for collaboratively training and deploying predictive models and integrate those models into real-time smart applications.
Start H2O Clusters
- Steam provides an interface to monitor memory usage and cluster performance. Increase resources without stopping or affecting tasks.
- Secure multi-tenant environment for data scientists. Multiple team members can work in parallel, request resources and return them to the pool once done without IT intervention.
- Bring new team members to speed with a single interface to get started. Works with all standard Hadoop flavors and all the popular cloud services. Provides easy portability without affecting users.
Save and Compare Models
- Model management interface that stores all historical models with training and validation metrics.
- Teams can compare the performance of models over various metrics and choose the champion.
- Model management is crucial to ensure model governance, to prevent the possibility of runaway or erroneous models going to production.
- The scoring server can deploy a model and expose a REST API that developers can ping to get predictions on new data.
- Elastically deploy your models in a scalable manner.
- Compile a model POJO to production
- Download a scoring service wrapper