Learn Amazon SageMaker – Second Edition: A guide to building, training, and deploying machine learning models for developers and data scientists

For data scientists who want to save time and money with AWS.
I have been following @Julien since 2007 who is the author of this book. He presented at numerous conferences and recorded hundred of videos at every level for everyone who has an interest in the subject. Today I am honored to review his book the second time which happened to be the second version of the book too. This book for myself is a note and reminder of the topics which I have seen and experienced so far. The author is extremely knowledgeable on not only SageMaker but other AWS services too. He manages to have all the available AWS certifications. Learning AWS Sagemaker from him is an amazing experience in form of the book.This book is exceptional when it comes to learning SageMaker, it starts with a clear beginner-friendly overview and SageMaker Studio which is the brain of this service in AWS.There are a total of thirteen (13) chapters in the book. The first 4 chapters are great for a beginner for who has less exposer in machine learning and wants to get hands dirty with starting with an overview of Service / Data Preparation using Data Wrangler / AutoML / Training Model with building in algorithms and basic model and deployment.Chapter 5 and 6 – Cover Computer vision and NLP which are hot topics today. On CV side Image classification, Object detection, and semantic segmentation are well explained. SageMaker and AWS have made these complex topics super easy. Instead of investing your months of time and energy now, it can be done within hrs. The author has touched upon every aspect of feature and associated services which are needed to cover these complex topics. This helps developers which have some AWS knowledge and coding experience can make an end to end projects in less time. NLP BlazingText, LDA, NTM are well covered in the book with examples. Chapter 7 – Covers built-in frameworks in Amazon SageMaker. Running your framework code on Amazon SageMaker. Using the built-in frameworks. Having Some knowledge of Docker is helpful. This is an advanced topic! The most interesting part is Hugging Face. The author himself working for hugging face now! Chapter 8 – Contains a lot of advanced info and a good understanding of Docker. Training and deploying with your custom Python code on MLflow. Building fully custom containers for SageMaker Processing etc.Chapter 9 – From this chapter onwards advanced training techniques have been covered such as Scaling training jobs SageMaker Debugger, pipe mode, distributed training, data parallelism,and model parallelism.Chapter 10 – This chapter covers managed spot training (50-70% $ saving), automatic modeltuning, SageMaker Feature Store, etc. Chapter 11 – Deploying Machine Learning Models (Inference pipeline, Multi-model Endpoint – “I used in my company”, Batch Transform, Model Monitor)Chapter 12 – Automating Machine Learning Workflows (AWSCloudFormation and AWS Cloud Development Kit (CDK), Step function stole my heart! The AWS step function is more powerful when you use with SageMaker it takes ML to next level with ease. Chapter 13 – Optimizing Cost and Performance – Autoscaling an endpoint, Deploying a multi-model endpoint, Deploying a model with Amazon Elastic Inference, Compiling models with Amazon SageMaker Neo.This book is helping me a lot in understanding how Machine Learning works at AWS and passing the certification exam also.I will highly recommend this book, 533 pages are well glued with amazing info.

Great Guide to Amazon SageMaker’s Ever-expanding Feature-set
Being such a large and ever-expanding ML platform, I find it challenging to keep up with the breadth of Amazon SageMaker’s features. Similar to the first edition, I found “Learn Amazon SageMaker (Second Edition)” to be adept at covering all the current features and functions of SageMaker in an easy-to-understand format for non-Data Scientists like myself.I also found significant value in the book’s focus on the general ML process independent of SageMaker – preparing data, building, training, deploying models, and automating your ML workflows.Lastly, since the cost of ML is frequently a concern of many organizations I work with, I appreciated the final chapter of the book, “Optimizing Prediction Cost and Performance.” The author claims prediction costs are “…typically accounts for 90% of the machine learning spend by AWS customers.”Disclosure: I received a copy of the book from the publisher for an honest review.

Well written end to end guide to AWS Sagemaker
Have to give it to the author for maintaining the smooth flow and simplicity while taking us through the seemingly niche topic of Machine Learning on AWS. After reading the book, no doubt you will end up learning the subject, but the book also provides value add with valuable information on using built in frameworks like Hugging Face, Apache Spark etc. As a solutions architect, my favorite section of the book is the one on optimizing cost and performance which is helping me incorporate these new learnings to my day to day job. The book makes the learning easy with attached screenshots from AWS console.