Google Expands Vertex AI MLOps with New Model Registry

Google rolled out a series of new features for its Vertex AI MLOps service on April 6.

Originally announced in May 2021, Vertex AI is a managed machine learning (ML) platform that is intended to help enable machine learning operations (MLOps). The new services that Google announced on April 6 include Vertex AI Model Registry and Vertex AI Workbench. The Model Registry is a central location for users to govern and manage the lifecycle of a given set of ML models, while Workbench provides a unified interface working with models across different tools in Google Cloud.

"Vertex AI is a managed platform that provides every ML tool that the customer needs to be able to build, deploy, and scale models," June Yang, vice president of Cloud AI and Industry Solutions at Google, explained during a media briefing. "We call this MLOps."

Vertex AI MLOps Brings Workbench to Users

The new Vertex AI Workbench service is intended to bring data and ML systems into a single interface.

The goal with Workbench, according to Yang, is to provide teams with a common toolset across data analytics, data science, and machine learning.

"This capability enables teams to be able to build, train, and deploy models five times faster than traditional machine learning notebooks," she said. "In fact, one major global retailer was able to draw millions of dollars in incremental sales and deliver a 15% faster speed to market with Vertex AI Workbench."

Related:How IT Operations Sustainability Efforts Can Make a Difference

Vertex-AI-Workbench

Vertex AI Workbench is directly integrated with Google's BigQuery cloud data warehouse technology as well as Google's Dataproc data query and Dataplex data fabric services.

Model Registry Accelerates MLOPs Workflows

Vertex AI also now helps organizations maintain high-quality models before and after the model is deployed, Yang said.

In an effort to help manage multiple models, Google has released the Vertex AI Model Registry service. Vertex AI Model Registry provides a central repository for discovering, using, and governing machine learning models, including those in Google's BigQuery ML, according to Yang.

Vertex-AI-Model-Registry

In terms of trends, Yang said that the concept of DevOps, DataOps, and MLOps are increasingly blurring and the new Vertex AI services are evidence of that fact. Organizations are trying to create applications and business processes that help inform the decision-making process and improve the overall end-user experience, she said.

"Because this functionality makes it easier for data scientists to share models, as well as developers to be able to use them, teams are ultimately more empowered to turn data into real-time decision making," Yang said.

Related:Top 10 Tech Buzzwords for IT Operations Teams in 2022