One of the frustrations with data scientists is that the models they develop do not always get used, and this is often due to a lack of collaboration with engineers and software developers, including within an organization. To address this, Tech Crunch reports that Rajen Sheth, who is the director of product management for Google Cloud’s artificial intelligence and machine learning products, aims to introduce mechanisms that will enable models to be more easily passed from data scientists to others so that applications can be developed from the data science models. These tools are designed to help data scientists to maximize their work.
These mechanisms are Kubeflow Pipelines and these are available for free an dare they are open-sourced. Kubeflow pipelines are an extension of Kubeflow, which is a framework built on top of Kubernetes, which are designed for machine learning. Pipelines are a type of containerized building blocks that data scientists can string together to build and manage machine learning workflows.
Quoted by Venture Beat, Sheth states: “One of the biggest problems we’re seeing right now is companies are now trying to build up teams of data scientists, but it’s such a scarce resource that unless that’s utilized well, it starts to get wasted.”
He adds: “One observation we’ve seen is that in probably over 60 percent of cases, models are never deployed to production right now. So we’re building a number of things to hopefully help cure that.”
Google Cloud is also to launch AI Hub. This is a location – a public repository – for data scientists to head to in order to find different kinds of machine learning content, which includes Kubeflow pipelines along with other modules. As well as accessing information, data scientists will be able share information for private purposes, accessible only to their own organizations.
Taken together, Google’s Kubeflow pipelines and AI Hub are meant to help data scientists share models across their organization. Hence, with Kubeflow Pipelines and AI Hub, data scientists’ work should be more accessible and more valuable to the rest of the organization.
Also announced by Google was the beta release of three new features in Google’s Cloud Video application programming interface, which is used to search and understand the content of videos.