SberCloud, a supplier of cloud services which forms part of the Sber ecosystem, has created a special hybrid version of its ML Space platform -  ML Space Private. This was announced by David Rafalovsky, Sberbank Group CTO and Executive Vice President, at the international conference AI Journey 2021.

ML Space from SberCloud is a platform for full-cycle ML development and joint work by DS teams on the creation, training and deployment of machine learning models. The platform contains tools that are familiar to data scientists and has pre-installed popular libraries and frameworks. It speeds up, optimizes and greatly simplifies model development, testing and deployment processes through the ability to use individual GPU V100 and A100 cards or the entire capacity of two supercomputers. 

ML Space Private offers absolutely all benefits of a public ML Space. Yet at the same time, it can be deployed in a hybrid regime – i.e., part of the platform can be installed on the client’s severs, while the other part performs computing tasks in the SberCloud cloud.

If required, ML Space Private can be certified to meet the information security requirements of any company.

Companies can also integrate the platform with their infrastructure and business services, such as service catalogs, billing systems, monitoring systems, logging, ITSM and so on.

The platform is popular amongst both Sber’s own ecosystem developers and data scientists in companies of any size and research organizations. In just one year it has been used by 1.5 thousand people from around 250 teams. In that time, they have used 2.5 petabytes of data and 300 mln GPU minutes.

David Rafalovsky, Sberbank Group CTO and Executive Vice President:

“SberCloud regularly receives requests from a wide variety of companies to work with the ML Space on-premises, because for one reason or another these companies are unable to use public cloud services. Of any providers – not just ours. Specifically for these clients we have created a special version of the ML Space platform – ML Space Private. I’m confident that this upgrade will open up broad opportunities to employ the most advanced ML technologies for many companies that have been unable to use the services of cloud providers due to regulatory or other restrictions.  Anybody who is interested can already submit an application to the website to test ML Space Private in operation”.