How to implement a standardized process in execution and delivery of data science solutions: Microsoft’s Team Data Science Process (TDSP)

To assist enterprise data science (DS) programs mature, Microsoft has proposed The Team Data Science Process (TDSP), which applies established software engineering practices to DS workflows. TDSP is designed to help DS teams with guidelines and tools to improve productivity and quality. TDSP provides a DS lifecycle process definition, standardized structure to organize projects, strategies for version control to promote collaboration and quality, and DS utilities to improve efficiency. We will present the TDSP principles and components, and how TDSP has been integrated within various DS and consulting services organizations who deliver advanced analytics services to customers.

Debraj GuhaThakurta is a Senior Data Scientist Lead in Microsoft’s AI & Research. His effort focusses on the use of different platforms and processes (such as Microsoft’s Azure ML, Cortana Suite, Cognitive Services, ML Server, SQL Server, Spark, Team Data Science Process), for creating scalable and operationalized AI solutions. Debraj has extensive industry experience in machine learning applications in biopharma and forecasting domains. He has a Ph.D. in chemistry & biophysics, and post-doctoral research experience in machine learning applications in genomics. He has published more than 25 peer-reviewed papers, book-chapters and patents.