Not All ML Heroes Wear Capes

Video size:

0:09 Miko Pawlikowski

Hello and welcome to another exciting episode of Conf42Cast. My name is Miko Pawlikowski and today my guest is Joshua Arvin Lat. He's the CTO of NuWorks Interactive Labs, previously three australian-owned companies, also CTO, AWS Machine Learning Hero and the author of the book "Machine Learning with Amazon SageMaker Cookbook" from Packt publisher. Hello Joshua, how are you doing?

0:36 Joshua Arvin Lat

Hello, good day. Great. I've been super busy this past couple of months. But yeah, I feel super great right now.

0:44 Miko Pawlikowski

Oh, Joshua is a serious man. He woke up at 4am for this today. So thank you so much, really appreciate that. So to warm you up, we typically start by confusing our guests a little bit. So the question for you is, what's your favorite planet?

1:01 Joshua Arvin Lat

Other people might look at, let's say, the other planets in the Solar System, or maybe other planets outside of the Solar System. But I think what we usually tend to take for granted, that would be our own planet, that is Earth. Because for one thing, I read somewhere that it's a living organism. So when we were starting to learn about the planets, what the living thing is, we always think of living organisms, as let's say, the humans, the animals. But I read somewhere that the Earth is a living thing. And one of the reasoning there would be that Earth, for some reason, is able to sustain life inside that atmosphere and that's one of the primary things that a living organism would be doing. So yeah, so if we don't take care of Earth, we also lose everything there inside. So in Earth, there's other things we can still discover about her. And yeah, that's the reason why it's my favorite, because people take it for granted, because we always start looking ahead too much, and doing that at other planets also.

2:18 Miko Pawlikowski

Love that. Definitely underappreciated. And I can't help but mention that, as you started talking about the living planet, that made me think about the "Guardians of the Galaxy" sequel and the Ego living planet. But I like your answer a lot. All right, Joshua. So obviously, I had to stalk a little bit online, well, check your LinkedIn. And the first thing that one notices when looking at your background is that you've had a series of startups when you were involved, mainly in the CTO facility. So I'm curious, you know, how it came to be? Did you just feel that, you know, entrepreneurial stuff is what you want to do? Was it just a series of things that you did with people that you liked? Or, curious, how does one basically do four different startups in a space of, what was it, eight years or something like that? Can you talk about that a little bit?

7:03 Joshua Arvin Lat

I started, I didn't really want to be a public speaker, didn't expect myself to have this kind of life 10 years ago. So this probably started when there was an international competition, when I was still in my last year in college. So we had this research paper, I think it's about two factor authentication system, a hybrid of that. And then there was competition, the first round was Asia Cup. And then I was the representative of my team from the Philippines. I think there are two teams from the Philippines. And then I was the representative there. And we had to defend our research paper in front of multiple judges. Yeah. And luckily, we won that competition. And it was like a 15 to 30 minute speech. So of course, I had to do improve my speaking skills and abilities, because you only had 15 to 30 minutes to present something that is very technical in nature, and something that is there to solve a certain problem. And then after a few months after winning the first round, in Asia, we had the International Cup, where it's international, meaning all the winners from the regional cups will be there to compete. So luckily, we won again, first place. Yeah, we were champions. And then, it was super tough, because everyone was super talented, their research papers were amazing, some were even I think, already doing Masters and PhDs. And we were just Undergrads back then. And we were aware that other countries have better educational systems also. So I mean, luckily, we won that. And after that, I got invited in Rutckon, one of the security conferences here in the Philippines, to do a talk about my research paper. So yeah, of course, when when joining those types of conferences, I knew that they were experts in the field, and I was just a student back then. So from my end, I knew my research paper well, and just shared it during the event. So yeah, that was my first experience. Those are my first few talks. And then what I did was, I tried to stop speaking for two years. So this is one of the stories I already shared in the past. And people will ask, why did you stop speaking for two years? Because I realized that sometimes when you always speak there's always a tendency for the human brain to think that you are already succeeding. But then there's that weird feeling that if you always speak, you will feel that you're already an expert, when in fact, your hands on skills and actual experience may be lagging a bit behind. So what I told myself is 'Okay, I'm starting to get this talks. I should I stop speaking first for, let's say, two to three years. And then I should focus first on my graph'. So what I did is at work, I spent, let's say, 9 to 10 hours at work when I was 21 years old, or 22 years old. And then after work, I would spend the next five hours there training myself. So it's like Batman training where I tried to upskill myself four to five hours per night after work hours, for two years. So after one year, I realized that a person's work, let's see, one week or two weeks, I would be able to do it in one minute, no, eight minutes, something like that. It did happen at work, where we had to install Redis. And then since I already did it before, all I did was install three commands, configure something really quickly. And then ta-da, you have your Redis server. Even at that point, I was still a frontend developer. And it was weird for me to know all those things, even though I am supposed to be doing frontend, not server-related stuff. So after realizing that, okay, my hands on skills are there, and I am already doing a lot of things. And I'm already comfortable telling myself, 'Okay, I'm an intermediate professional. I'm no longer someone who's just good at public speaking', I decided to go back and start speaking again. I wanted to start from scratch and didn't want to use my early success in the global competition as a way to shortcut a bit and tell everyone, 'Hey, I'm an expert'. I wanted to start from zero again. And yeah, I realized that it plan out well, because after my first talk, the second talk came after, and then third, fourth. And then right now, after there's a certain year, I think, three years or four years ago, where I had about, let's say one talk per month. And then this last couple of years, there were some weeks where I had four talks in one week. So it's super crazy. And from my end, I realized that when I'm sharing my knowledge to the different people, it helps remove the stress, it helps me relax and rest. I guess for one thing, when you're sharing your experiences, it helps people that's number one. And number two, they ask questions, they ask for advice. I get to meet people outside my company, and then I get to solve their problems also. So usually, when you use a certain tech stack, for example, you only are aware of how it's used in your current company. But when people ask you genuinely for help, and you're there to help them, you're able to gain a lot of things. For one thing, you become more of a natural leader, because for one thing, you become generous in sharing a solution to other people. And then you also get to analyze their own situation. And that will bring both people growth, in a sense. And yeah, so I learned a lot of things along the way and public speaking is something that was able to help me connect with other people, help them. And until I became a Machine Learning Hero at some point, which I was surprised at, I became machine learning hero. Probably due to the public speaking engagements.

13:14 Miko Pawlikowski

I definitely agree that genuinely helping someone grow and learn something new is probably one of the more satisfying experiences that are there to begin with, and seems to be working well for you. So for everybody who's not familiar with what it actually means to be AWS Machine Learning Hero, it sounds a bit funky. What does it mean?

18:50 Joshua Arvin Lat

Given that I wrote the book for about six to nine months, the feeling was kind of different every single month. Yeah, for one thing, I had a very busy schedule at work. So I had to squeeze in the time writing the book. Yeah, so to share some context, I'll share the experience of writing the book each month so that at least you have a better understanding of what's happening each month. So before writing the first chapter of my book, I did a bit of research first. So what I did was I opened amazon.com and then looked at each book and then looked at the comments section. And then looked at the comments and ratings with the lowest ratings. So there's one to five stars ratings there, right? So, I had to look at what makes people hate a certain item or book. I listed those down. And then for one thing, one of the common issues when reading a book is that customers hate it when you try to copy-paste content online and put it in their book. Because of course, there's going to be redundancy there. If you're going to spend money on a book, why would you buy a book if it's already there, if you can already find it online? So that's one example. And sometimes the code samples are not even working when you purchase the book. Or maybe there's a lot of explanation I'm skipping inside the book. So after doing maybe half a day of research on what makes a good book good, I started writing the first chapter. Writing the first three chapters was easy, because when you're writing a book, you already have an idea, especially in the first parts. And towards the middle of the execution part, what happened there was AWS started releasing new services. That's the tricky part, because I was already done with about 50% of the book. And then, for some reason, about October, November last year, AWS had reinvent. Every year, AWS announces a lot of new features and announcements. And topic that I'm writing on, which is SageMaker, had about four or five new features, major features. It's like a feature, but if you dig deep into those features, you'll realize that it's like a completely different service inside SageMaker. So for one thing, let's say that you want to do ML Ops inside SageMaker without using another service, SageMaker had SageMaker pipelines. And I had to learn all of those things, even if it's really new, because I had to include it in my book, along with the other enhancements in the existing services. So, doing a bit of research and revising some chapters is one of the trickier parts that I had to deal with. But it's fun, because I wanted to make sure that the quality of the book is super high. Because right now, I was aware that during the pandemic, people definitely had to worry about a lot of things. And, let's say, if you were watching a movie, you would probably know that the quality of movies during the pandemic, the quality is probably lowered a bit during the pandemic, because it's much harder to work this past two years. So, one thing that I told myself, 'Okay, it's better to write a book that is super high quality', and I had to spend some extra time enhancing the quality. And yeah, it turned out well. And the funny thing there was, November last year, the book already had about 1000 pages. November, December last year, we had about 900 to 1000 pages. And then the Packt team told me, 'Sorry, you can only do 600 to 700 pages'. I was like, 'Wow, what will happen to the extra 300 to 400 pages of content?' And after researching a bit and trying to understand what we were saying they told me that, 'You cannot do anything more than 700 pages because it might affect your book'. Because if you were to print it, the publishing, the printing press shop or something would not be able to print something that's more than a certain number of pages. Other ways the book might be in unstable size, something like that. So yeah, I had to trim down the content from 1000 pages to 700 or 600. And that's one of the trickier parts there. So whenever there's something like that when doing something, had to take a step back, rest for the night, and then try to solve it again the next day. Because of course, the feeling is tough, when the things that you work on would just be thrown away, right? Especially the extra pages of work. So yeah, that's probably one of the things that I experienced. And then I decided to think positively and tell myself, 'Okay, one of the definitions of perfection is that something is perfect, if you can no longer take something away from it'. Yeah. So for example, if you have a tool, you can remove the fluff in order to improve its quality even further. So what I did is I reviewed the entire book, and decided to reduce the pages from 1000 to 700. And the removal of the 300 pages there, after reviewing it multiple times, did improve the book, because it removed the fluff, it removed the redundant explanation. It removed the redundant content. And the things left there, is something that is really concise and straightforward. Because sometimes when you're writing a book, there's a tendency to write multiple things that share the same concepts. So at least there, I am really sure that this 760 page book has all of the contents there that's unique and there's no redundant information. And some things there's a page there, 'Please refer to this certain chapter or recipe for an explanation, because it's explained there in detail'. And then towards the end, towards the end, especially this past couple of months, it's more on the marketing initiatives reaching out to, let's say, editorial reviewers. Because when you have a book, the books have something like editorial reviewer section, where they get to read the book first. And then they give one to two sentence of comment, let's say, 'Okay, this book extensively discussed the contents, the features of SageMaker' or something like that. And then yeah, I believe it will be released October 29. There are ups and downs, but mostly ups.

25:45 Miko Pawlikowski

Mostly ups.

25:46 Joshua Arvin Lat

Yeah. Mostly, when you're doing something that is like a passion project, you want something to be as perfect as possible, something that if you were to read it a few years back, this is the book that you want to read, so that you will learn something the fastest time possible. Because when you're trying to learn machine learning right now, and trying to apply it in the cloud, there's some sort of disconnect. Because some of the tools that we're using right now may probably work super fast in local machines. But when you have to do it in the cloud, there are very few references about it, especially on the topic of applying machine learning. So machine learning, yeah, you can learn, there are these datasets, and so on. But once you have to deploy it online, there are a few references, there are a few good references. But there are really few references on the topic of let's say, SageMaker, applying machine learning with SageMaker, using different libraries and frameworks with SageMaker. So, let's say you're using TensorFlow, Scikitlearn, Pytorch, usually you have your own books for those topics. But trying to migrate it, or port it, to SageMaker, there's very few of that topic. And also trying to do things end-to-end in the cloud, I also included there in the book, along with the troubleshooting techniques. Because that's something that's also hard to find online. Sometimes when you're trying something out, it's hard to solve something because even if you look online, given that there are very few answers online, I had to simplify it and make it something that's easily consumable by someone who would read it. So whether you're a beginner, an intermediate user, or an advanced user, you would appreciate it in the book. And to share a story also. When I was doing something at work, when we were trying to prepare a SageMaker Notebook instance. And we were trying to use Athena. So Athena as the service for doing queries on big data. When we were doing that at work, I completely forgot how to configure something in Athena. So what I did is I opened my book, because I have a copy of my own book before it gets published. So I opened my book and used the step-by-step guide there on how to configure things properly, especially when connecting to your S3 Bucket, which would serve as some sort of database, where you can do queries automatically, just reads. So yeah, that's one of the advantages of writing your own book, you can also use it as your own reference, when you're doing stuff related to it.

28:08 Miko Pawlikowski

Okay, so one last question. For everybody who's kind of in a similar situation, they've been thinking about writing a book, their first book, or maybe they were approached, what's one piece of advice that you would give them?

32:00 Joshua Arvin Lat

Mr. Joshua Lat.

32:05 Miko Pawlikowski

There you go, you can follow the man. And for everybody listening, we've got a giveaway. We've got five books from Joshua, "Machine Learning with Amazon SageMaker Cookbook" to give away. All you have to do to get one of this is to send us an email with Joshua in the title to mark@conf42.com. And the first five people are getting the book. For everybody else, I'm afraid they're going to have to be satisfied with a discount coupon. Thank you so much for your time, Joshua. This has been a pleasure and all the best with the sales of the book.