A great resource about Machine Learning Ops, the new DevOps for Machine learning pipelines. Check out our review!
Introducing MLOPs: The following is a review of the book Introducing MLOps: How to Scale Machine Learning in the Enterprise by Mark Treveil & the Dataiku Team.
Review of Introducing MLOps
We’ve said it a million times in reviews of other books like ‘Building Machine Learning Powered Applications‘ or ‘Data Science in Production‘: the theory and fundamentals of Machine Learning and Artificial Intelligence are beautiful, but the truth is our beloved Boosting or Deep Learning models are no good if they’re not deployed in a production environment where they can be used to provide real value.
Because of this, it is starting to be essential that Data Scientists feel comfortable abandoning their intuitive and visual Jupyter notebooks, and learning how to face a Machine Learning project using proper development methodologies.
For traditional software development these methodologies were known under the name of DevOps or Development Operations. In the data world, they’re known as MLOps or Machine Learning Operations.
What is MLOps?
MLOps is a set of practices that aims to build, deploy, and maintain Machine Learning models in productions reliably and efficiently. It seeks to standardise and improve the procedures that make turning a data set into a Machine Learning model embedded in an application or in real use case.
It divides this whole path into a set of steps like data collection and processing, feature engineering, data labelling, model design, training and optimisation and finally model deployment and monitoring.
Most Data Scientist only tend to cover the feature, labelling, and model building phases, feeling rather uncomfortable outside these areas. Because of this, more and more data scientist, in order to improve their skills and opportunities are learning about data and software engineering, becoming ML engineers or full stack data scientist.
By learning MLOps you will learn how to do this, and what you should be doing at every step of the process.
Introducing MLOps is a very brief introduction to the world of MLOps, aimed at those Data Scientist that know how to clean and prepare data and build models that want to learn how to turn these models into precious value.
This book introduces the key concepts of MLOps to help data scientists and application engineers not only operationalize ML models to drive real business change but also maintain and improve those models over time.
Through lessons based on numerous MLOps applications around the world, nine experts in machine learning provide insights into the five steps of the model life cycle–Build, Preproduction, Deployment, Monitoring, and Governance–uncovering how robust MLOps processes can be infused throughout.
This book will help you in the following areas:
- Achieving full data science value by reducing friction throughout ML pipelines and workflows
- Refining Machine Learning models through retraining, periodic tuning, and complete remodeling to ensure long-term accuracy
- Design the MLOps life cycle to minimize organizational risks with models that are unbiased, fair, and explainable
- Operationalize ML models for pipeline deployment and for external business systems that are more complex and less standardized
Despite promising a lot of content, for us this book is too brief (less than 200 pages) and its content could be summarised in even less. The material is very introductory and there is no deep insight into any topics like CD/Ci for Machine Learning.
If you want a quick Introduction to MLOps then this text can be appropriate, however, if you really want to learn and feel more comfortable with your Machine Learning pipelines you will need additional material. Don’t worry, we got you covered:
- The Coursera MLOps Specialization: a good online resource that covers how to conceptualize, build, and maintain integrated systems that continuously operate in production. Covers more content than Introducing MLOps.
- The MadewithML site
- The ML Tools Landscape with almost 100 Machine Learning tools – interesting if you know MLOps and want to find new softwares/practices
- MLOps website: This for us is the best free resource with tons of high quality material.
Summary & conclusion on Introducing MLOps
In this post we’ve reviewed what MLOps is and the book ‘Introducing MLOPs’. We would not buy it for the money it costs, seeing the large range of free materials out there, however if you want a lightweight introduction to MLOps in paperback, to carry around, then this migth be the resource for you. You can find it on amazon here:
- Treveil, Mark (Author)
- English (Publication Language)
- 183 Pages - 01/05/2021 (Publication Date) - O'Reilly Media (Publisher)
As always we hope you liked the review. If you have any comments or want us to review a specific text, send us a message to firstname.lastname@example.org. Also, don’t forget to follow us on Twitter and have a great day!
Tags: Introduction to MLOps, What is MLOps, Machine Learning in production