Vertex AI, Google’s managed machine learning (ML) cloud platform to help organisations accelerate the deployment and maintenance of artificial intelligence (AI) models, is out .
Announced at the Google I/O developer conference, Vertex AI is described as “a unified data science and ML platform for all skill levels”. It brings together Google Cloud services for developing ML under one unified user interface and API in order to simplify the process of building, training, and deploying machine learning models at scale, according to Google.
However, the Google Cloud release notes reveal that Vertex AI is a new name for the AI Platform (Unified) that Google launched last year, but with additional features and extensions.
The aim of Vertex AI is to provide a single environment through which organisations can actually see models through from experimentation all the way into production. With this, data scientists and ML engineers across all levels of expertise should be able to implement Machine Learning Operations (MLOps) and build and manage ML projects throughout the entire development lifecycle.
“We had two guiding lights while building Vertex AI: get data scientists and engineers out of the orchestration weeds, and create an industry-wide shift that would make everyone get serious about moving AI out of pilot purgatory and into full-scale production,” said Andrew Moore, Google’s vice president and general manager of Cloud AI, on the Google Cloud Blog.
Google states that Vertex AI is based on knowledge the cloud giant has gained from building and deploying ML models in production. As a consequence, users will have access to the same AI tools used by Google in-house such as computer vision, language, conversation and structured data.
Vertex AI introduces features such as Vertex Vizier, which optimises a model’s output by tuning the hyperparameters automatically, a fully managed Vertex Feature Store to allow developers to share and reuse ML features, and Vertex Experiments to accelerate the deployment of models into production with faster model selection.
To support MLOps, tools such as Vertex Model Monitoring, Vertex ML Metadata and Vertex Pipelines are available, which streamline the end-to-end ML workflow, according to Google. Meanwhile, Vertex ML Edge Manager provides the ability to deploy and monitor models in edge deployments with automated processes and flexible APIs.