Amazon Web Services is constantly driving new innovations that empower data scientists through the use of a vast number of machine learning cloud services. But what does it take to effectively handle machine learning projects and implement and evaluate algorithms on the cloud?
Mastering Machine Learning on AWS, set to be released on June 11th, has the answers.
We sat down with authors Saket Mengle and Maximo Gurmendez for a behind-the-scenes look into the coming of Mastering Machine Learning on AWS.
Q: Congratulations!! What an amazing accomplishment for both of you. When did you first realize you wanted to write a book?
SM: Writing a book was always on my bucket list. I met the publishers at one of the conferences I was speaking at. When they approached us with a proposal to write a book, I could not pass on such an opportunity.
MG: I’m a fan of learning from books because it provides a comprehensive approach to a particular topic. I like to read every bit of books, the preface, the author bio, the acknowledgments, etc. So the idea of writing a book was always something I wanted to do. When Saket told me about the opportunity I was ecstatic.
Q: Who is the book geared for? Who should read it?
SM: I have been working in the field of machine learning for more than 13 years now, and have conducted numerous interviews for the position of data science. My objective for the book was that if a person reads this book and finishes all the exercises, they should be able to pass the interview for a data scientist position with flying colors. The book is geared towards students and professionals who want to transition to the field of data science.
MG: This book is part of a mastering series, so it does assume some basic knowledge on the subject, mainly college math and basic python programming. Ultimately, I think this book is targeted at doers. By doers I mean those that want to learn how to leverage machine learning concepts and lots of great technologies, like AWS, Spark and Tensorflow to build a smart product. The book is not a comprehensive guide for each of the technologies, mostly tries to show how all these tools can be tied together to create amazing things with little effort.
Q: What are the top 3 things readers can expect to learn from the book?
SM: The book covers all popular machine learning and deep learning algorithms. Our aim is to provide easy to understand examples to the readers that would help them learn complex machine learning algorithms. We also wanted to provide an explanation to show what makes each algorithm unique and useful in specific situations. The book also covers coding examples on how to implement each algorithm in Apache Spark and Amazon Sagemaker.
MG: I hope that through this book readers will learn 1) when to use which algorithm for which business problem, 2) how to improve their models, both from the data science as well as the engineering perspective, and 3) how to productize machine learning to release smart product features on their businesses.
Q: What is your work schedule like when you’re writing? How do you fit it in and still find time to enjoy the writing part?
SM: I mostly wrote the book during nights and weekends. The initial chapters were difficult. I experienced every symptom a writer faces when writing a book. However, with each chapter, the process became easier as we received really good and prompt feedback from our publishers on our chapters.
MG: As Saket says, we had a pretty clear plan. The schedule was a bit crazy, as we’re both very busy, but I wrote mostly on weekends and late at night when the kids were asleep. I spent a good portion of my time working on selecting datasets and crunching data on jupyter notebooks to be able to tell a good data science story, relevant to the topic at hand. Sometimes our notebooks would not work as intended or did not achieve the result we were aiming for so had to iterate several times.
Q: How long did it take to you to write the book?
All: We took 6 months to write the book and an additional month to work on the review process with our editors.
Q: What would you say is your interesting writing quirk – now that you are a famous author?
SM: I realized that I was able to write at a faster rate when I was tired, compared to when I was well-rested.
MG: We had set deadlines for each chapter, which sometimes meant taking advantage of every little moment. I remember writing Tensorflow portions of the book while spending the day on my brother-in-law’s boat, in Punta del Este.
Q: Where do you get your information or ideas for your books?
SM: I brushed up on each of the algorithms using my old textbooks before I started writing the chapters. However, my aim was to come up with easy to understand examples to explain each algorithm. Hence, I tried to take inspiration from the real world applications where each algorithm can be applied.
Each of our chapters not only presents the algorithm, but also a real-world application area where we show its application. Our objective is that the readers would be able to apply the machine learning algorithm in their business applications after reading the book.
MG: I also used different sources, from traditional academic text books, to blogs written by many authors. That said, I mostly took the ideas from similar problems that we had to solve in the past.
Q: What do you like to do when you’re not writing?
SM: I like to watch movies and play video games when I am not writing.
MG: That’s my problem. I like to do a lot of different things. Mostly, I like to sing, play the piano or guitar, write songs, and enjoy recording them on my home-studio.
Q: What was one of the most surprising things you learned in creating your book?
SM: Writing a book was always something I wanted to do. However, my first concern was how I would get time to write the book. I learned that if you are passionate about getting something done, then you always find a way. I would have to thank my wife here, who let me spend hours in my home-office and supported me through the process.
MG: While digging deeper on some of the topics, I learned how hard it is to do some of the most basic things when it comes to some of the newer technologies and frameworks. I’d say it’s mostly due to the absence of up-to-date documentation. For example, to deploy a Tensorflow model through SageMaker I had to go and actually read the source code of each of these two products. We hope this book helps users overcome such barriers, through end-to-end practical examples we made available.
Q: Where can we find a copy?
Q: What’s next? Any more books?
SM: I am going to enjoy the next few months to reflect back at the writing process of this book and the feedback I receive. However, I enjoyed the writing process greatly, and I am sure I will be back to writing another book when I find the right topic.
MG: Same thing! I need to get my weekends back now and spend quality time with my family. However, I really enjoyed the experience and hope to be able to come back to writing in the near future!
Thank you so much for your time and best of luck with your new publication. We are proud to share this news with the dataxu community!