I first met the author, Ville Tuulos, in 2012 when I was trying to understand the hype around Hadoop. At the time, Ville was working on Disco, an Erlang-based solution to map-reduce that made it easy to interact with Python. Peter Wang and I had just started Continuum Analytics Inc., and Ville’s work was a big part of the motivation for releasing Anaconda, our distribution of Python for Big Data.
As a founder of NumPy and Anaconda, I’ve watched with interest as the explosion of ML Ops tools emerged over the past six to seven years in response to the incredible opportunities that machine learning presents. There are an incredible variety of choices and many marketing dollars are spent to convince you to choose one tool over another. My teams at Quansight and OpenTeams are constantly evaluating new tools and approaches to recommend to our customers.
It is comforting to have trusted people like Ville and the teams at Netflix and outerbounds.co that created and maintain Metaflow. I am excited by this book because it covers Metaflow in some detail and provides an excellent overview of why data infrastructure and machine learning operations are so important in a data-enriched world. Whatever MLOps framework you use, I’m confident you will learn how to make your machine learning operations more efficient and productive by reading and referring to this book.