Atos has made available a software suite intended to make it simpler to build artificial intelligence (AI) applications using a combination of intellectual property developed by Atos and open source components that have been vetted by the global systems integrator.
Jerome Sandrini, head of Big Data and cybersecurity for Atos North America, says the Atos Codex AI Suite is an application toolset that Atos has built around machine and deep learning algorithms that makes it simpler to build AI applications.
“We’re using best of breed components,” says Sandrini.
The basic idea is to create a framework that makes it simpler for teams of developers and data scientists to collaborate on the development and training of AI models. Atos Codex AI Suite provides access to development tools, an orchestration software, a deep learning engine, and a repository through which data scientists can easily share code and models. In fact, Sandrini notes that Atos Codex AI Suite includes pre-trained algorithms that organizations can reuse to develop those AI models.
The most challenging and time-consuming aspect of building AI models is not the development of the core application, but rather the effort required to train algorithms to identify and react to patterns and anomalies. In fact, over the course of the lifecycle of an AI application, it’s probable that multiple models will need to be trained as more data concerning how a process or task is supposed to optimally function becomes available.
The Atos Codex AI Suite represents an extension of other application toolsets that Atos makes available to enable customers to reuse code and processes it has developed over the course of multiple IT services engagements. That code is designed to be deployed across public clouds and local servers as well as platforms residing on the network edge, says Sandrini.