Skip to main content

Components

H2O eScorer consists of three components:

  1. eScorer service on HAIC
  2. Wave app on the App Store
  3. Python client

eScorer service​

With both Fully Managed and Hybrid Cloud, deployment options, H2O eScorer is a production-grade model deployment and scoring service that is secure and auto-scalable with high availability.

The following is a list of supported model types:

  1. H2O Driverless AI MOJOs
  2. H2O-3 open source models
  3. H2O Python Scoring Pipeline (for use cases like time series)

H2O eScorer Wave app​

App Store on the H2O AI Cloud provides state-of-the-art AI web apps, including the H2O eScorer.

Note

For more information, see Access H2O eScorer.

The H2O Wave app connects to the eScorer backend and automatically initializes your models, deployments, and other metadata. Once you're logged in, you can use the interactive GUI for model deployment, model stats, model management, and lifecycle - all in one app.

Note

For more information, see H2O eScorer flow.

H2O eScorer Python client​

H2O eScorer comes with a lightweight, low-latency Python library with robust APIs for your enterprise applications. These APIs are simple yet feature-rich, and bring you the most effective way to productionize your H2O.ai models for real-time and batch scoring.

Note

For more information, see the Python client guide.

H2O AI Cloud​

The H2O eScorer Service and the Wave app run on the H2O AI Cloud. Among the various benefits of HAIC's world-class, state-of-the-art AI/ML workflows, the key advantages include the flexible deployment options:

  1. H2O AI Hybrid Cloud: complete control over infrastructure, software updates, security, and compliance. Customers can operate the Hybrid Cloud in any cloud or on-premises IT environment.
  2. H2O AI Managed Cloud: security, infrastructure, and high availability. This means that we manage your infrastructure provisioning and scale the cloud computing capabilities as per your needs, while you focus on what's most important for your business.
Feedback