Ville Tuulos has spent more than two decades developing tools and infrastructure for data science and machine learning. His career has been driven by a single question: how do you make data scientists as productive as possible, from quick local prototyping all the way to production workloads that scale reliably?

A centerpiece of that work is Metaflow, the full-stack framework for data science that Ville designed and built at Netflix. Metaflow grew out of the realities of running data science at Netflix scale, where projects need to move from a laptop experiment to production without friction, and it is now an open-source framework used to structure real-world ML and data science workflows.

Today, Ville is the CEO of Outerbounds, a startup focused on data science infrastructure that builds on his experience with Metaflow and large-scale ML platforms. He shares what he has learned in his book Effective Data Science Infrastructure, which walks practitioners through designing infrastructure that serves data science, rather than forcing data science to serve infrastructure.