Joseph Sirosh, Microsoft | BigData New York City 2014

Joseph Sirosh, Microsoft, at BigDataNYC 2014 with theCUBE's Dave Vellante and Jeff Frick

Microsoft endorses value of the app economy | #BigDataNYC

https://siliconangle.com/2014/10/22/m...

The rise of the app economy has created a unique opportunity for developers to pursue their ambitions, producing multi-billion dollar success stories such as Instagram and WhatsApp, but the data science community does not yet have a medium through which to channel its potential. Microsoft Corp.’s Joseph Sirosh dropped by theCUBE at SiliconANGLE’s recent BigData NYC meet-up to share how he and his team are changing that.

As the head of the machine learning group within the Redmond giant’s booming cloud business, Sirosh is at the forefront of the effort to bring about what Cloudera Inc. chief office strategy Mike Olson described as the “Big Data app economy” back in 2011. For the vision to come together, he believes that the data science scene needs its own version of the App Store.

“Data scientists don’t build software apps, they build machine learning models, they build analytical systems, they build visualizations – they don’t have an outlet to monetize this,” said Sirosh, who has himself been practicing the trade since before data became a buzzword. Microsoft hopes to provide that outlet in the form of the Azure Marketplace, which allows users to publish pre-packaged algorithms and machine learning models for others to consume.

The strength of that model comes not from groundbreaking analytics projects but rather the much more common everyday use cases such as production recommendations on retail websites, Sirosh stressed. That kind of data-intensive task not only requires the know-how of a data scientist but also the scalability that Microsoft promises to bring to the table with Azure.

“Building these recommendations is a hard job; it takes a data scientist to do that. But there’s another part that’s equally hard: To build an API that hooks into that web page, handles the volume of traffic and reliably serves recommendations with every click,” Sirosh explained. The last ingredient in his vision for the Big Data app economy is domain expertise, an invaluable asset in addressing analytics challenges limited to a specific department or industry.

Traditionally, addressing such a specializing need required companies to hire their own data scientists, who are scarce and expensive. Sirosh compared the challenge to custom-tailoring.

Thanks to automation, however, “Manufacturing clothes became so easy and so automated that you can go to a department store, select the sort of things you like and buy it,” he added. “So you need to create something like that: finished services that I can buy.”

Microsoft is aiming for exactly that with with its Marketplace, which, continuing the clothing analogy, Sirosh described as an effort to build a “factory in which data scientists create a large number of intelligence services so that you can get your APIs and just go.”



@theCUBE #BigDataNYC ‪@Microsoft‬ #theCUBE ‪@siliconangle‬ #BigData