Cortana Analytics Suite Powers Russell Reynolds Associates’ Search for the Perfect Match

This post is co-authored by Ke Huang, Data Scientist, Said Bleik, Data Scientist, and Tao Wu, Principal Data Scientist Manager, all at Microsoft.

Russell Reynolds Associates, a global leader in assessment, recruitment and succession planning for C-suite roles, CEOs and boards of directors has teamed up with Microsoft to develop an innovative search system that pairs ideal candidates with the right positions for them by tapping into more available data and by applying learning during the search process.

Conventional recruitment searches can be labor intensive, requiring manual efforts that rely heavily on information stored on databases and searched using handcrafted queries. Using Microsoft’s big data and advanced analytics capabilities, Russell Reynolds Associates now can search structured and unstructured data as the system learns to identify “good matches” with user-labeled data. Mining data that include resume text and metadata (high-level structured data on a candidate’s experience), as well as position description text and metadata (high-level structured data on a position), Russell Reynolds Associates can quickly find the best candidates for specific positions. Machine learning models are built and deployed as a web service interface on Azure ML with high availability and scalability that meets Russell Reynolds Associates’ needs.

Russell Reynolds Associates’ search tool is implemented on Cortana Analytics Suite using two systems – for structured and unstructured data. Within the Azure ML portion of the suite, a series of data preprocessing and feature engineering procedures is performed on data stored on Azure SQL Database. To identify ideal candidates, a binary classification model is developed to determine the probability of a candidate and a position being an ideal match. A matching decision then can be made based on a chosen decision threshold. To help find additional candidates and positions, a measuring array for similarities is constructed based on either processed structured data or unstructured data. A threshold then is selected to help filter out irrelevant candidates or positions based on Russell Reynolds Associates’ requirements.

“We are extremely excited to be able to offer our clients the latest in machine learning and data analytics,” says Mark DiStaulo, Global Head of Application Development at Russell Reynolds Associates. “This system, utilizing Microsoft’s unique capabilities, will allow us to identify and pair the best talent with search assignments, leveraging data to which our competitors do not have access. This is one of several capabilities we intended to deliver as parts of our innovation roadmap.”

End-to-end data pipeline using Cortana Analytics Suite and Azure.

The Azure Data Factory component of the suite is used to schedule periodic data transfer from on-premise SQL Server to Azure SQL Database automatically, as shown. When data becomes available on the Azure SQL DB, it gets processed, and Russell Reynolds applications can consume prediction results using the Azure ML web service interface. These results also are copied into Azure Blog storage.

Using the fully managed big data and advanced analytics capabilities of Cortana Analytics Suite, Russell Reynolds Associates is able to build intelligent applications that better match senior C-suite executives with their ideal future careers.

Tao
You can contact Tao at tao.wu@microsoft.com.