Linux company Red Hat Inc. is trying to encourage more people to create their machine learning models on its OpenShift platform with the launch of its new Red Hat OpenShift Data Science service today.

Red Hat OpenShift is a Kubernetes platform that helps companies orchestrate large clusters of software containers that host the components of modern applications. With OpenShift, companies can create, modify and deploy applications on demand, enabling rapid software development and release lifecycles.

Red Hat OpenShift Data Science is a fully managed cloud service for OpenShift that promises to deliver an advanced yet simplified machine learning experience on the OpenShift platform.

The idea, Red Hat said, is to provide data scientists with a more customizable machine learning platform compared to existing alternatives that restrict them to using a very specific set of tools. Red Hat OpenShift Data Science boasts an open workflow platform that includes Jupyter notebooks and common frameworks such as PyTorch and Tensorflow, and is complemented by access to certified partner technology from the Red Hat Marketplace.

It can be thought of as a fully integrated machine learning model development environment that provides easy access to various useful tools from third parties.

So, for example, users can access services that make it possible to optimize and tailor their machine learning models’ behavior on Intel Corp. hardware using the Intel OpenVINO for Enterprise toolkit. Then, they can access the Intel oneAPI AI Analytics Toolkit, which provides various tools and frameworks needed to extract insights from models and optimize their performance on Intel central processing units.

That way, the company added, data scientists have integrated access to the tools and frameworks needed to build and deploy their model as well as help in delivering high performance on Intel hardware.

Another key launch partner is Nvidia Corp., whose graphics processing units serve as an ideal engine for more demanding AI and machine learning models. Red Hat said its integration with Nvidia’s hardware and services will enable accelerated computing, giving data scientists the ability to scale computationally expensive neural networks to large, complex architectures without sacrificing productivity. In other words, data scientists will be able to reduce the time it takes to train their models by throwing more resources behind them, with minimal coding changes required.

Other integrations include Anaconda Inc.’s popular data science platform; IBM Corp.’s Watson Studio with AutoAI, which makes it easier to build, run and manage AI models at scale; Seldon Technologies Ltd.’s Seldon Deploy tool, which can be used by data scientists to refine their ML deployment models; and Starburst Data, Inc.’s Starburst Galaxy platform for data analytics.

“For many companies the biggest barrier to adoption is the complexity of wiring together the necessary data sources with diverse model training and model deployment technologies,” said Mike Piech, vice president and general manager of cloud data services at Red Hat. “With Red Hat OpenShift Data Science we’re helping organizations overcome such complexity to begin harnessing the full potential of machine learning from the leader in trusted open source technology.”

Constellation Research Inc. analyst Holger Mueller told SiliconANGLE that data science is one of the major use cases for implementing new platforms, so the new service has potential to grab the attention of more enterprises.

“What is key is that providers offer solutions that are turnkey, as much as possible, because speed is critical for enterprises looking to create next-generation applications infused with AI and machine learning,” he said. “This is what Red Hat gets right with today’s offering, which bundles more than just software, with lots of useful partner offerings too.”

Red Hat said Red Hat OpenShift Data Science is available now as field trial add-on to Red Hat OpenShift Dedicated and the Red Hat OpenShift Service on Amazon Web Services. The release includes general availability-level support, meaning users can test the service while only paying for the underlying infrastructure that powers it.

Image: Red Hat