MLOps with Databricks

Foreword by Alejandro Saucedo

Today, many teams can build models and many can build platforms. However, far fewer can connect those two disciplines with the clarity and depth required to build platforms and engines that let organizations continuously deliver value with high speed, reliability, and trust.

Maria has spent years doing exactly that. Her work across machine learning operations (MLOps), platform thinking, and now large language model operations (LLMOps) reflects not only her technical fluency but a grounded understanding of what it takes to move from experimentation to reliable, repeatable delivery. As I have had the pleasure to know Maria for many years, I can confidently say that she belongs to a rare group of practitioners who can bridge the gap between two things that often remain divided in the field of machine learning: the rigor of software/platform engineering for production ML and the flexibility to empower scientists with the ability to deliver business value. Today, bridging that gap matters more than ever.

When Maria mentioned that she was going to consolidate all the knowledge she had gathered in a book, I was very excited. As I have been able to read the multiple iterations of this book, I can say that it does what it initially set out to do. It does not treat MLOps as a checklist of processes or the Databricks platform as a collection of isolated features. Instead, it approaches both from first principles. Maria breaks down how real production excellence ...