Google Cloud Vertex AI: New Managed ML Platform

Google Unveils Vertex AI: A New Machine Learning Platform

During Google I/O, Google Cloud introduced Vertex AI, a newly developed, fully managed machine learning platform. This platform is designed to streamline the deployment and maintenance of AI models for developers. The announcement at I/O, traditionally focused on mobile and web development, underscores the significance Google places on this new service for a broad spectrum of developers.

Addressing Challenges in Enterprise Machine Learning

The development of Vertex AI stemmed from a thorough evaluation of the current landscape within Google Cloud. According to Craig Wiley, Director of Product Management for Google Cloud’s AI Platform, machine learning in the enterprise is facing a critical juncture. He highlighted that numerous reports and analyses indicate a substantial gap between investment in machine learning and the realization of tangible value.

Image Credits: Google

Wiley, formerly the General Manager of AWS’s SageMaker AI service, observed that while organizations like Google have successfully leveraged machine learning’s transformative potential, initial cloud service offerings often involved a proliferation of services, some of which proved unproductive. The primary objective of Vertex AI is to accelerate the return on investment for enterprises, ensuring they can derive genuine benefits from their machine learning models.

A Flexible Platform for Model Training and Management

Vertex AI is engineered as a highly adaptable platform. It aims to empower developers and data scientists of all skill levels to efficiently train models. Google claims it requires approximately 80% less code to train a model compared to certain competitors. Furthermore, the platform facilitates comprehensive management of the entire model lifecycle.

Image Credits: Google

The service integrates seamlessly with Vizier, Google’s AI optimizer. This integration automates hyperparameter tuning in machine learning models, significantly reducing tuning time and enabling faster experimentation.

Features for Enhanced Collaboration and Deployment

Vertex AI also incorporates a “Feature Store” to facilitate the serving, sharing, and reuse of machine learning features. Additionally, Vertex Experiments are included to expedite model deployment and selection for production environments.

Deployment is supported by continuous monitoring and Vertex Pipelines, a rebranded version of Google Cloud’s AI Platform Pipelines. This feature assists teams in managing the workflows associated with data preparation, analysis, model training, evaluation, and production deployment.

Multiple Interfaces for Diverse User Needs

To cater to a wide range of developers, the service offers three distinct interfaces. These include a drag-and-drop tool for ease of use, notebooks for advanced users, and, notably, BigQuery ML. BigQuery ML allows users to create and execute machine learning models using standard SQL queries within Google’s BigQuery data warehouse.

“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,” stated Andrew Moore, Vice President and General Manager of Cloud AI and Industry Solutions at Google Cloud. “We are very proud of what we came up with in this platform, as it enables serious deployments for a new generation of AI that will empower data scientists and engineers to do fulfilling and creative work.”