Although this book bears our names as authors, many other people contributed to its creation. Without their help, this book wouldn't exist, or at best would exist in a lesser form. Kenyon Brown was the acquisitions editor and so helped get the book started. Christine O'Connor, the managing editor, and Caroline Define, project manager, oversaw the book as it progressed through all its stages. Sonam Mishra was the technical editor, who checked the text for technical errors and omissions—but any mistakes that remain are our own. We would also like to thank Matt Wagner from FreshBooks, who helped connect us with Wiley to write this book. Finally, we would like to thank our wives for their patience as we spent many weekend hours researching the content and writing this book over the past 8 months.
About the Authors
Shreyas Subramanian has a PhD in multilevel systems optimization and application of machine learning to large-scale optimization. He is currently a principal machine learning specialist at Amazon Web Services, and he has worked with several large-scale companies on their business-critical machine learning and optimization problems. Subramanian is passionate about simplifying difficult concepts within optimization, and he holds two patents in areas connected to aviation-related tools and techniques for improving efficiency and security of the airspace. He has also published over 20 conference and journal papers on the topics of aircraft design, evolutionary optimization, distributed optimization, and multilevel systems or systems optimization. He has several years of experience building machine learning and optimization models for customers in large enterprises to small startups, while taking part in and winning hackathons on the side. Subramanian is passionate about teaching practical machine learning to citizen data scientists and has trained hundreds of customers in private, hands-on environments and has helped customers build proofs-of-concept that are now in production today, providing millions of dollars’ worth of revenue to the AWS business as well as customers.
Stefan Natu is a principal machine learning (ML) architect at Alexa AI, where he is building an ML platform for Alexa scientists and engineers. Prior to that, Natu was the lead ML architect at Amazon Web Services, where he focused on financial services and helped major investment banking, asset management, and insurance customers build and operationalize ML use cases on AWS, with an emphasis on security, enterprise data, and model governance. Natu has developed and evangelized common ML architecture and infrastructure patterns globally across AWS highly regulated customers, leading to numerous production ML deployments and millions of dollars in AWS cloud revenue. He has authored over 25 AWS machine learning blogs, code samples. and whitepapers, and is a frequent speaker at conferences such as AWS re:Invent. He completed his PhD in atomic and condensed matter physics from Cornell University, and he worked as a research physicist at ExxonMobil, submitting two patents and over 25 peer-reviewed publications. Natu is passionate about mentorship and has served as a technical adviser at Insight Data Science, where he guided students in their transition from careers in academia to industry.
About the Technical Editor
Sonam Mishra is an IT consultant with several years of experience in diverse roles ranging from software development, to application testing, to technical content creation. She is passionate about new and emerging technologies, particularly in the area of cloud computing. She lives in the United Kingdom with her family.
Introduction
Machine learning (ML) is one of the most popular and rapidly growing fields in the technology industry today, with far-reaching business implications. The market for ML solutions and products is expected to grow annually by tens of billions of dollars, and with it, the demand for professionals who understand how to analyze data and build ML solutions is expected to grow as well.
ML is a highly technical field, and successful ML professionals need a foundation in mathematics, statistics, and data analysis. They must be able to code and have a fundamental understanding of infrastructure and software development best practices. In the past, the practitioners of machine learning were academics and PhDs, but the industry demand for ML is much larger than the supply of new PhDs emerging from academic institutions.
The purpose of this book is for you to understand the concepts and principles behind ML, with the practical goal of passing the AWS Certified Machine Learning Specialty exam. As practicing ML solution architects, we go well beyond the scope of the test in this book and incorporate architecture patterns and best practices that we have seen employed in the industry today. Reading this book will also give you an understanding of what is required to be a successful machine learning architect.
This is not a book on ML foundations. That is simply too vast a field for us to do it justice in this book and also is not our intention. There are a number of excellent textbooks and online resources you can use to develop a foundation on ML algorithms, deep learning, and similar topics. However, we will cover the concepts that you will need for the test.
Finally, one of our favorite leadership principles here at Amazon that widely applies to the solution architect role is learn and be curious. We have found that the best way to learn a topic is to get hands-on, and we highly recommend that you go beyond this book and get hands-on experience in ML. Download and explore some public datasets, and train some simple predictive models. Build a neural network from scratch using TensorFlow/PyTorch or just native Python. Explore AWS services such as Amazon SageMaker by running some of the sample Jupyter Notebooks. We highly recommend getting some hands-on knowledge before taking the test. Check out the AWS Training and Certification web page for helpful courses: www.aws.training.
Don't just study the questions and answers! The questions on the actual exam will be different from the practice questions included in this book. The exam is designed to test your knowledge of a concept or objective, so use this book to learn the objectives behind the questions.
The ML space is maturing and growing very quickly; what this means is that our book is just a snapshot in time of our understanding of the industry and certification requirements. We highly recommend that you read the SageMaker home page to review the latest releases that may appear on the test.
The AWS Certified Machine Learning Specialty Exam
The AWS Certified Machine Learning Specialty exam is intended for professionals who perform a data science, machine learning engineer role. The official details of the test can be found here: https://aws.amazon.com/certification/certified-machine-learning-specialty.
The focus of the test is to validate your understanding of foundational ML concepts, foundations of statistics, data analysis, exploration, feature engineering, and common ML algorithms. This is required knowledge for anyone performing this role in industry today. However, in addition to this, this certification focuses on your ability to deploy those solutions on AWS and to be able to architect an end-to-end solution on AWS from data ingestion to model deployment and monitoring using a host of relevant AWS services for a given business use case.
Why Become AWS Machine Learning Specialty Certified?
There are several good reasons to get your AWS Certified Machine Learning certification:
It provides proof of professional achievement.
