 Hello. My name is Andrew Ho. I am a product manager and I'll be talking about breaking into AI ML product management. Yeah, so let's begin. So currently I am a product manager at Meta AI, previously known as Facebook. I'm giving this talk because the primary motivation for me is to share my experience and how I broke into product management for artificial intelligence and machine learning. And this is not so much of how to break into AI ML, but more of one path that was taking my personal path and that I've taken. And just sharing it to everyone so that everyone has an idea of what path can be taken. And then I'll share some general guidance of if you are interested in breaking into the AI ML space, what are some things that you can do, what are some resources that you can tap into to at least take you a few steps closer there and help you towards that eventual goal of reaching and breaking into artificial intelligence and machine learning product management. So for the agenda today, we'll be covering a brief background of myself and then we'll talk about my personal product journey and I'll talk throughout how I went from Microsoft to Airbnb to Meta. And then finally I'll conclude with some of the lessons that I've learned and hopefully that's useful for other folks as well. So about me, my product journey began at Microsoft where I was a PM on Microsoft Azure from there, I transitioned over to Airbnb. And I was Airbnb, that's where I really broke into the machine learning space. And then finally, I am currently a product manager at Meta, formally known as Facebook, working on a number of cross org product management machine learning initiatives. So what is AI ML product management? At the very fundamental kind of definition, it's building products that either use your produce machine learning models, but it goes a lot deeper than this. It further suck segments into two different aspects. One is your applied machine learning, where you're using different machine learning models to enhance a product's capability or feature, or it might even be the core capability of the product. And there's also the platform building or building the infrastructure for how machine learning products are built. I belong to the latter, but I can definitely chime in and give some insights towards the former as well. Beyond that segmentation, it goes even further, where there is different categorizations for specialization. And this tends to differ depending on whether you're at a big company or a small company. Small companies tend to rely more on specialized product managers, while bigger companies can take more generalized product managers. And these product managers might look to specialize later in their career. So what are some of the specific examples of AI ML products? So one of the first ones is, this is from Meta, where when Facebook and Ray Bands did a collaboration to release Ray Bands stories, some of the key capabilities include object detection within these augmented reality glasses. And so you see that machine learning is actually taking a role to enhance a product in the physical world. And object detection can be used for many different reasons, such as detecting vehicles, oncoming traffic, enhancing what you actually see from these glasses. Next, we dive into the virtual world, where one of the common problems within the financial industry is using classification models to detect credit card fraud. This is an ever going battle between developing machine learning models that are better at detecting fraud and fraudsters for improving the ways so that they can bypass these machine learning models. A great way of binding machine learning and application machine learning with the finance sector. And lastly, it's building the infrastructure to build and manage and deploy models. This is definitely my bread and butter. This is the specialization that I went into. A lot of the machine learning models today are built off of these type of platforms and building these platforms help companies scale and help them manage all of the different use cases that they have for machine learning. So I encourage everyone who is listening to this webinar to really think about what type of machine learning that you're really interested in and what is the motivation for getting into AI ML product management. Are you interested in building ML features into these type of augmented reality devices? Are you interested in finance and how that applies or healthcare? Are you interested in building the infrastructure or the cloud space? That initial motivation that you have, that insight will be greatly useful to help guide and be the compass in just how you develop your skillset and which way that you want to proceed. So my product journey begins at Microsoft. I joined Microsoft to work on Azure Cosmos DB. Azure Cosmos DB is Azure's premier NoSQL distributed database. And this is a time when the cloud was early on. Some of the core foundational service was built and more of the past offerings were starting to be developed. And so AWS had the S3, the EC2, while Azure had Azure Blob Storage, Table Storage, Compute Network. But some of these newer past offerings are starting to develop and the cloud space was growing much, much faster. And so my goals when joining Microsoft and in particular Azure was to really hone in my product management skillset. And so I wanted to really refine what does it mean to be a product manager? What are these skillsets? And I'll cover a few. They include your customer acquisition, customer empathy parts of it. It includes your product roadmaps, product vision, the execution, how to actually build a product, go to market strategies, as well as a number of other things as well. Next, I wanted to build my technical acumen. Joining a really technical product such as a cloud offering within Microsoft was quite a challenge. I wanted to really challenge myself and build this technical skills where I can be a product manager that can work really closely with engineering and really figure out how things work, and also build these for potential customers who are engineers who use Azure as the backbone for their infrastructure. And then finally, another one of my decisions, which I encourage folks to also think about is who are you going to work with? You are, it is very cliche, but you are going to work with these folks every day, nine to five. And so it's worthwhile to really admire and want to work with these folks. And so I've had the opportunity to work with an extremely talented team of product managers and engineers where I felt that most of these individuals could be my mentors, and I've actually learned a tremendous amount. And my time within Microsoft cemented my product management experience, which I leveraged today and transitions through today. And so I'll talk about some specific points of the things I've learned while working at Azure Cosmos DB. So first, it's a go-to-market strategy. At a very high level, go-to-market strategies are always about how do I have the biggest splash when I launch my product while making sure that I mitigate any risks so the product doesn't flop. And so that includes defining your minimum viable product. What are the minimal viable product requirements that we can make to launch? And how do you make those tough decisions of whether we can cut a certain feature or not to make sure that we hit a certain deadline or timeline? Pilot customers, extremely tricky because you are looking for customers who will take the first big bet on your product. And you're looking for customers who will give you insightful feedback so that if your product is missing some or having some large gaps, they'll tell you and they'll tell you that it needs to be fixed. And so you can make sure that you build the best product before we launch. Support channels. How do you have the right open, free support channels so that people can communicate if things are not going right? Documentation, press release, which is mostly around the marketing piece. How do you make the biggest splash when you actually go out with your product? Next, the other critical learning that I got from my time at Azure was building infrastructure at scale. And this really helped build my technical expertise and my learnings around distributed systems. This is the domain that I first entered in. And this is where I wanted to really hone in my skills as the technical product manager. And I saw the opportunity to do so. And so I am incredibly thankful for my time at least at Microsoft to learn about all the technical skills that I've developed. And then finally, going from zero to one, I was fortunate enough to join a team where we were just starting to build out a product. And that meant that we had to go from zero to one. We had to chase our first internal customers, chase our first external customers. We had to creatively solve challenges because oftentimes you're taking up multiple hats and multiple roles to make sure that you can grow this product. And the other key piece is cementing your core value proposition because this product is going to last for a while. And your core value propositions of what your product delivers and how the internal team and how yourself view the product is going to last for a long time. And so making sure that's really solidified. So these are some of the few things that I've learned at Microsoft and there's a whole other laundry list as well. But I just wanted to highlight these. And so if I learned so much in Microsoft, why did I actually leave to Airbnb? And how does actually play a role in how I got into the AI ML space? It was a chance of fate to be quite honest. During that summer, I broke my wrist while playing basketball with some friends. And that really motivated me to think about my next step in my product journey. I was in bed recovering from the surgery. And I really wanted to think about what do I want to do next? What are the skill sets that I want to develop? And towards the end of my time within Azure, I really started to think about the ML space and how Spark ML connected with Azure Cosmos DB and how maybe I can start diving into the ML space. And so what did I do next? I'm stuck in bed. I'm recovering from the surgery. I am doing the most that I can to try to make lemonade from lemons. And so I've started to think about how do I dive into the ML space? First, it's, I know that I had a really strong technical background working at Azure. I saw that these can definitely transfer and not only the technical parts, but just the core product management learnings that I've had from Azure could transfer as well. Next, how do I start diving into machine learning? How do I really kind of understand a new, relatively new domain to myself? And so this is where I started studying ML in my spare time. I've built machine learning models to get familiar with the domain. I've listened to, I can't say how many podcasts of machine learning to really familiarize myself with this space. I knew that I'm still a novice, but I definitely took the first couple steps at this time. And I really tried to upload myself as much as possible to learn about machine learning and learn more about the foundation of machine learning. So lucky enough with all the learnings that I've got, as well as the learnings I've had from Azure, as well as the ones that I've learned being stuck recovering. I was able to go to Airbnb and be the product manager for their machine learning infrastructure, their feature engineering framework, which is called Zipline and the former called Big Head, and eventually their labeling teams, generalized labeling team, as well as the applied machine learning teams. So what did I learn at Airbnb? What was my biggest learnings as a product manager at Airbnb? The first was about machine learning platforms. I was really able to combine my experience within Azure with a new domain of machine learning to really understand what does making a machine learning platform entail. And so it's the infrastructure for training model development, feature engineering, model and feature inference or model and feature management and inference systems. Next, this was a big change. I went from an external product. So Cosmos DB was external, it was intended for external customers to onboard. While this was an internal product, this was going to be used by solely Airbnb teams. So what were the differences here? How do we set OKRs? How do we acquire buy-ins from all the different stakeholders, from all the different orgs within Airbnb? How do you do road mapping? How do you alignment? This was quite a change for myself. And one of the other lessons I wanted to highlight here is open source technologies. I got involved in the open source community. I wanted to understand this new and thriving community within machine learning. I've met with a number of experts throughout the community to learn how Google, how Netflix, how Lyft, how Uber, how all these other companies are doing machine learning. And I was able to understand the startup seen as well. What are startups doing in machine learning? How are they revolutionizing machine learning platforms? And so my learning really continued while I was at Airbnb. And this takes me to the final step of my journey as of today, which is currently I am at Meta. I work on their machine learning platform across a number of orgs. I drive product for a number of products there. And my learning continues to this day. I'm learning more about cross org product vision. How do you define a vision that spans across so many different orgs and how do you make sure that they're all aligned and they're all pushing with 100% effort towards this vision? How do you scale machine learning? Airbnb scale is quite large, but Facebook slash Meta is significantly larger. It is the machine learning and the problems to solve there is quite interesting at the scale that Meta is at. And so how do you build platforms for some of the largest models in the world? And one of the things I really admire the new skill set that I'm starting to acquire is communication. For such a large company, concise and clear communication is extremely important and how you share your messaging and how you make sure that people are onboarded and how many people are bought into your vision and they understand what you're going to work on. And that includes getting buying from a number, a large number of stakeholders. So that was my personal product journey on how I got into AI ML. There are, if I had to distill this down into three key lessons that I encourage everyone to really think about is the first and number one most important thing is do not hesitate to begin building your AI ML skill set. The amount of material out there is so vast that if you choose to take the first step, there is so much material out there for you to learn from. I've put in a few things here in case people are interested, but you can build your first model. You don't need to pay any money. These are all free and these are really easy to use. You can start with Kaggle. They have a Jupyter notebook or notebook environment where you can actually build your first model there, test it out, try it out, see how you like it. You can also go to Google CoLab or you can use any of the free trials from the core cloud providers out there such as AWS, GCP or Azure. But I really encourage you to build your first model. The first model, you don't need to creatively think about how you'll solve it. You can even follow a tutorial. But just understanding the steps of building your training data, the feature engineering, the actual model development and how you get your predictions is really crucial. I highly encourage it. It doesn't take that long. You can definitely go through a tutorial in just a few hours. That's my number one recommendation. Next, taking intro course to machine learning. There are so many good ones out there. I've taken a number of myself. I recommend one of the fundamental ones that's been there for a while, but it's probably the most popular. You've probably already heard of it, which is the Corsair Intro to Machine Learning by Andrea. I highly recommend it. It's a great way to start getting started into machine learning, start understanding some of the basics around it. There's also a Udacity course for Intro to Machine Learning. Third recommendation, watching webinars, ML talks, virtual conferences. After you watch this one, I'm hoping you're watching this one. You can go and watch other products, cool talks. I have linked a few out there that I watched and are quite good. And so that's another way of really getting started. And the last one, although it might seem difficult, I do encourage you to speak with experts. There are a lot of experts out there who like to talk about machine learning. I personally love to talk about machine learning, and I love to talk to folks who are willing to also talk about it. And it doesn't really matter if you know quite a lot about machine learning or you're just beginning. It's about having a conversation of learning. And so highly recommend it. Feel free to reach out to these folks. Feel free to reach out to me on LinkedIn. I'm always happy to talk, happy to really help talk about more about my experience or more about the AI ML space. And if you are looking to build your first model, some of the recommendations I have, the first one is predicting housing prices with linear regression. This is the beginning of machine learning. This is the one that everyone starts with. And so it's your linear regression. It's not the fancy neural network yet, but this is where it starts and this is the heart of it. And so I recommend that. And then the second one is the PyTorch convolutional neural network with the endless dataset, extremely famous dataset for figuring out how do you actually get these images of these digits and how do we classify it as zero through nine? And so extremely, extremely, two extremely great ways of getting started building your first model, hopefully soon. Next, leveraging your current strengths as a product manager. So when I talked about my product journey, I really talked about how every time I had an opportunity to work at a phenomenal company, I learned so much. And I want to emphasize that breaking into AI ML product management, the first steps you've probably already taken, it's just building a really solid core product management skill set. And so you've probably already built customer empathy, product vision, clear communication, how to write great PRDs and lean into your strengths. I highly encourage this where when I transitioned from Azure to Airbnb, I heavily relied on the skill set that I developed in Azure to break in. And so I recommend that to you as well, lean into your technical expertise, whether you're really good at sales and marketing, vision, execution or strategy, lean into it because these all play a tremendous value. And then the AI ML part will start to come. As you work in the field, you'll start to learn it, but heavily lean into the skills that you already have and lean into your strengths. And finally, embrace the challenges and don't be afraid to try and fail. Oftentimes, I hear that it's intimidating to get into the AI ML space. It seems futuristic. It seems like what does this even mean? It's hard to understand where I can provide value. And so I highly encourage folks to embrace the challenge and not be afraid to fail because it's better to try and fail than anything else. And it's not supposed to be easy. And so failing is parts of it. I failed many times. And I'm sure I think all the other product managers I know in the AI ML space have failed many times as well. And so be bold, pursue the opportunities, incorporate the machine learning you're building as well as your original skill set, hone in on your craft. And I promise you things get easier. The first step is always the hardest step. And then the last one is you really don't have to wait for a broken risk to start something new. And so mine is a chance of fate that I broke my wrist. And that gave me the opportunity to break into this space. But I guarantee there's an easier way. You do not need to break a bone. You can get started way easier than that. Cool. And last, I do want to do a quick plug where Meta AI is hiring product managers. And so if you are interested, feel free to connect with me or you can apply directly. Both the links are there. And so thank you so much. Thank you for listening.