Using Jupyter at scale within an enterprise: 10,000 users, 10,000 notebooks

The National Security Agency (NSA) has a large use-case for Jupyter. While traditionally known as a data science platform, we have worked to expand Jupyter's use within our organization to include our large user-base of intelligence analysts. These analysts, hard at work culling and correlating information from a multitude of repositories, are the domain experts who are closest to the analytic challenges we face. Jupyter has empowered this community - who do not traditionally come from software engineering backgrounds - to translate their tradecraft into code, making that tradecraft more reproducible, more efficient, and more sharable.

The scale of our effort has forced us to address a number of challenges. With over a thousand Python authors, ten thousand notebooks, and ten thousand Jupyter users, we developed tools and approaches to manage, curate, and sustain crowd-sourced development of Jupyter notebook based analytics. We identified appropriate training paths to introduce Python and Jupyter into communities that most often lack prior backgrounds in coding. We outlined common use-cases for Jupyter, and worked to lower the barrier to entry to make the platform as approachable as possible. Most importantly, we ensured that all users of Jupyter within our Agency meet our strict data security and access controls.

This talk sheds light upon the secretive world of the NSA and shares some unique insight from our experience of adopting Jupyter across a large organization. We will also detail how we are working to support the open source community surrounding Jupyter.