Abstract: 

In this talk we discuss the fundamental differences between how data scientists and machine learning engineers approach the world, and how the resulting tension between them leads to one of the main barriers that must be overcome for ML projects to succeed. We describe how a machine learning platform like Kubeflow can dramatically change how these groups work and collaborate. We present several example scenarios in the lifecycle of a machine learning project, and show how an ML platform increases collaboration, reduces friction, and improves the lives of data scientists and ML engineers. Platforms, and the change in methodology that they enable, can dramatically transform culture, reduce risk, and help improve ML project outcomes.

Bio: 

Souheil is the Head of Field Data Science at Arrikto where he helps build machine learning solutions for clients. Previously, Souheil worked at Freddie Mac and Capital One where he built models and machine learning platforms. Prior to becoming a data scientist, he spent 15 years in academia working on MRI and Brain Imaging. Souheil holds a BS and PhD in Physics from Yale and MIT respectively.