News

Google's New Vertex AI Platform Enables MLOps

• By John K. Waters
• 05/19/2021

Google unveiled a new managed machine learning (ML) platform this week during its annual I/O conference, held online again this year. Vertex AI, now generally available, was designed to allow data scientists and ML engineers "across all levels of expertise" to implement Machine Learning Operations (MLOps) to build and manage ML projects throughout the development lifecycle.

MLOps is an emerging set of management practices for the production ML and/or deep learning (DL) lifecycle. It combines DevOps, ML, and data engineering processes to improve communication between data scientists and the operations or production team. Google claims the new platform requires nearly 80% fewer lines of code to train a model than competitive MLOps offerings.

Google designed Vertex AI to bring together the Google Cloud services for building ML under one unified UI and API, which simplifies the process of building, training, and deploying ML models at scale. From this environment, customers can move models from experimentation to production faster, more efficiently discover patterns and anomalies, make better predictions and decisions, and generally be more agile in the face of shifting market dynamics, the company says.

The marque feature of the new platform is its ability to allow developers and data scientist across skill levels to train ML models quickly without formal ML training. It comes with an AI toolkit used internally at Google that includes computer vision, language, conversation, and structured data, which is "continuously enhanced by Google Research."

With another standout feature, Vertex Vizier, Google provides a fully managed Vertex feature store, an ML-specific data system that stores and manages feature data, allowing users to serve, share, and reuse ML features. Another feature dubbed Vertex Experiments is designed to accelerate the deployment of models into production with faster model selection. And two MLOps tools, Vertex Continuous Monitoring and Vertex Pipelines, help to streamline the end-to-end ML workflow.

"We had two guiding lights while building Vertex AI," said Andrew Moore, VP and GM of Cloud AI and Industry Solutions at Google Cloud, in a statement, "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. 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."

During his part of the conference-opening keynote, Kemal El Moujahid, product director for TensorFlow and ML at Google, cited Vertex AI beta partner, Portal Telemedicina, a Brazilian healthcare startup that delivers low-cost diagnostics to more than two hundred cities in Brazil and Africa with an AI-assisted diagnostic service solution. The company used Vertex AI to develop models for scanning thousands of ECGs for heart problems across Brazil and Africa to help physicians better detect and triage high-risk patients.

In a press release, Google cited ModiFace, a part of L'Oréal and a market leader in augmented reality (AR) for the beauty industry. ModiFace is using the Vertex AI platform to train its AI models for such new services as skin diagnostic, which is trained on thousands of images from L'Oréal's Research & Innovation, the company's dedicated research arm. "Bringing together L'Oréal's scientific research combined with ModiFace's AI algorithm, this service allows people to obtain a highly precise tailor-made skincare routine," the company says.

The Vertex AI platform is generally available today.