Sergey - stock.adobe.com
Google Cloud debuts Vertex AI managed ML platform with MLOps
Users can build, test and deploy machine learning models and put them in production faster with Google Cloud's Vertex AI platform, which contains new and existing Google AI tools.
With its new Vertex AI managed ML platform, Google Cloud said it has boosted the ability of data scientists and engineers to use MLOps to much more quickly build, deploy and maintain ML models -- essentially turning machine learning into a production line-like process.
The tech giant released the Vertex AI system into general availability today, on the first day of the Google I/O virtual conference.
The platform unifies Google's disparate AI technologies -- including the AI toolkit that powers Google with computer vision and language, conversation and structured data -- with a series of new MLOps features under the Vertex name.
The vendor said that with the new managed ML platform it is responding to enterprises' needs to move from experimentation to production faster. With Google Cloud AI services, users can build ML models under a unified user interface and API.
“With Vertex AI we don’t need to build [machine learning models] ourselves,” Kemal El Moujahid, director of product management at Google – TensorFlow, said today during a keynote session at the conference, which is aimed at developers and runs until May 20. “Vertex AI will take care of that.”
Google deepens AI footprint
Google's latest AI offering appears to cement the vendor's position as the tech giant most clearly associated with AI, and the one with the biggest arsenal of AI and ML tools, said Chirag Dekate, an analyst at Gartner.
Being the major cloud services provider most identified with AI could also help Google -- now third in market share -- make inroads on No. 2 Microsoft, Dekate said. As of the start of 2021, AWS led the cloud services race with about 33% market share, with Microsoft at 18% and Google at 9%, according to Synergy Research Group.
Chirag DekateAnalyst, Gartner
Meanwhile, Vertex AI addresses a critical need facing enterprises as the COVID-19 pandemic has spurred a dramatic uptake of AI amid a profusion of technologies from vendors, Dekate said. Enterprises, however, are still stuck with many time-consuming ML processes that prevent most models from ever entering production, he said.
"They're launching an AI factory here, an AI industrialization platform that enterprises can customize and leverage," Tekate said. "The core goal is to accelerate the velocity of models."
Vertex AI essentially adds on to Google's old AI Platform collection of autoML services features sought by data scientists, such as experiment tracking, a feature store and autoML tables, said Kjell Carlsson, a Forrester analyst.
New MLOps tools
Google Cloud also buttressed its existing collection of autoML tools with new MLOps features such as Vertex Continuous Monitoring, what Carlsson called "an incremental improvement" on its ML model and maintenance capabilities.
"It's good -- don't get me wrong -- but they had a good amount of that before." He said. "Maybe now they're making it easier to consume and use. It's not a revolutionary capability, but it's still very, very important."
What's most notable in this area is the Vertex Pipelines feature, which provides much-needed connectors between processes, Carlsson said.
"Having that managed and governed all centrally such that the data engineer can do machine learning end-to-end, or the data scientist can do data engineering and they can collaborate on the same platform," he said. "That's something I haven't seen from anyone else."
New MLOps tools in Vertex AI include Vertex Vizier, which boosts experimentation speed; the Vertex Feature Store to enable users to serve, collaborate on and reuse ML features; and Vertex Experiments to accelerate introduction of models in production by enabling users to select models faster.
Google Cloud said several enterprise users have already been working with Vertex AI.
ModiFace, a unit of L'Oreal, the makeup and cosmetics giant, is using the system to train its ML models on thousands of images from L'Oreal's research division for new services such as enabling consumers to try beauty products using augmented reality.
Essence, a data-driven global media agency, is using Vertex AI to enable its developers and data analysts to update ML models to meet changing marketing needs.