Keynote: Data Science at Netflix: Principles for Speed & Scale
9:55 - 10:40 | Thursday, May 23
Data science powers Netflix. It informs our decisions and challenges our assumptions. It fuels experimentation and innovation at unprecedented scale. It helps us discover fantastic content and deliver personalized experiences for millions of people around the world. In short, data science has become critical to Netflix, touching nearly every aspect of the business today.
But it hasn’t always been that way.
Let’s take a quick step back and acknowledge a universal truth: data science is hard. So hard that Gartner estimates 85% of all projects will fail. The great irony perhaps is that it’s the ‘data’, not the ’science’, that troubles most projects. From missing and invalid data to bad training sets and incorrect interpretations, there are countless things that can go wrong—and frequently they do. Even relatively smooth projects can take months to go from ideation to running in production. And the bumpy ones? Those can take a year or more to complete, if they finish at all.
That used to be true at Netflix too. But now it’s possible to go from ideation to production in a fraction of the time.
In this talk, Michelle Ufford will discuss the philosophies and innovations that have made this possible. She’ll walk through what the process looks like today and how Netflix is tackling some of the biggest challenges facing data scientists. Through it all, she’ll share core principles that have contributed to their success, which you can put to immediate use at your own company.
Head of Data Science Tools, Netflix
Michelle Ufford leads the Big Data Tools engineering team at Netflix, where she’s responsible for platform innovation and usability tooling for Netflix’s industry-leading data platform. Some of her team’s current projects include Jupyter notebooks, workflow orchestration, platform alerting, data quality, and machine learning. Prior to that, she led data engineering, data management, and platform architecture for GoDaddy, where she set a TPS record for SQL Server and helped pioneer Hadoop data warehousing techniques. Michelle is also a published author, patented developer, and award-winning open source contributor. She currently lives in the San Francisco bay area and can be found on Twitter at @MichelleUfford.