Make with H2O AI: Overview of H2O AI Cloud
- Predictive AI
- H2O Driverless AI
- H2O Document AI
- H2O Feature Store
Overview
This is the first session in H2O's "Make with H2O AI" webinar series, giving a broad overview of H2O AI Cloud (now the H2O Sovereign AI Platform). Reid Maloney and VP of Products Vinod walk through the make-operate-innovate framework, covering H2O-3, Driverless AI, Hydrogen Torch, and Document AI for building models, and H2O Wave for building apps on top of them, then MLOps and Feature Store for managing everything at scale. A live demo shows an end-to-end workflow from a telco churn dataset to a deployed model.
Key takeaways
- How H2O organizes its AI Cloud around three stages: make (build models and apps), operate (deploy, monitor, govern), and innovate (share and reuse via an app store)
- How H2O-3, Driverless AI, Hydrogen Torch, and Document AI each target different problem types, from open source distributed ML to no-code deep learning and intelligent document processing
- How H2O Wave lets teams build interactive, real-time AI applications in Python without separate front-end development, shown through a hospital occupancy simulator and a transaction fraud review app
- How the Feature Store lets data science teams share and reuse engineered features across projects, illustrated with an AT&T example
- How H2O MLOps handles model registry, deployment, scoring, and monitoring for drift, bias, and other operational metrics across frameworks beyond just H2O
- A live end-to-end demo: importing a telco churn dataset, training an AutoML model in Driverless AI, reviewing Shapley value explanations and a bias check, registering the model, and deploying it as a real-time endpoint in MLOps
Start owning your AI