Microsoft's Project Sage: A cloud-based recommendation service

A Microsoft incubation project focused on building a machine-learning-powered recommendation service is staffing up and taking steps toward going commercial.
Written by Mary Jo Foley, Senior Contributing Editor

Microsoft is forming a new team that is building a recommendation-as-a-service, according to a new job posting.


The Microsoft Cloud and Enterprise Engineering team in Israel is building what it has codenamed "Project Sage." The job post describes Sage as a "Generic recommendation as a service solution that enables customers to use state of the art recommendation engine in a very easy and intuitive way."

That's it as far as official specifics as to what this recommendation service will do. I'd bet at least one potential use will be a recommendation service for publishers, similar to what Google is doing with its beta content-recommendation offering. I'd think an ad recommendation/personalization offering based on top of this is probably likely, as well.

Project Sage got its start as an incubation project two years ago and "is making its first steps as a product group," the job post says.

Project Sage has its roots in "CloudML," or cloud machine-learning. There's a subset of that team that builds algorithms that can be used by Microsoft product groups and external facing products. (The CloudML Algorithms team is also part of Microsoft's Cloud and Enterprise team, which is led by Executive Vice President Scott Guthrie.

From a separate job posting for an opening of the CloudML Algorithms team:

"Our team is unique in combining engineering excellence and research depth. The large number of internal users of the CloudML TLC toolkit provides for interaction with many products and learning about the wide variety of modern ML applications and techniques - from random forests on MapReduce to deep learning on GPUs, with everything in between."

Machine learning is one of foundational technology areas that are building blocks of Microsoft CEO Satya Nadella's cloud first/mobile first strategy. Cloud-based machine learning is also a big priority for Microsoft Research.

At the company's New England R&D (NERD) center, making machine-learning cloud technology accessible to enterprises, data scientists, developers, information workers, consumers and devices is part of the mission statement. The idea is to provide all users with access to a machine-learning platform that allows researchers to conduct experiments and developers both inside and outside Microsoft to build web services that are powered by the cloud. 

Microsoft researchers NERD and Microsoft's New York research lab have been working in conjunction with the Bing team on using CloudML to do predictive analytics. And on May 13, NERD is hosting its third annual New England Machine Learning Day.

Speaking of predictions, Microsoft officials announced on May 8 that they have added forecasting (and hindcasting) capabilities to PowerBI for Office 365, which is Microsoft's business-intelligence service.

(Thanks to @h0x0d on Twitter for the pointer to the job listing for Project Sage.)

Editorial standards