 AI product manager became the hottest job ever recently in the tech industry and tons of people want to break into product management as AI PM. However, most of them was left out in dark because the roles and responsibilities of AI product manager is very different than traditional product management. And I have personally launched award-winning AI products and also have helped many people learn the AI product manager job and some of them then had a plot of position in an AI company getting paid over $400,000 per year. In this video, I'm going to demonstrate the day in the life as AI product manager and how do we train AI model and how do we work with AI engineers so that you're able to see exactly what they look like as AI product manager so that you're able to break into the AI space very easily. Hey guys, this is Dr. Nancy Lee, a director product and featured in Forbes. I've helped 100 people learn the dream PM job offer in fan companies and unicorn startup and continue to get promoted as a product leader. In this channel, we cover free product management training and tech trends. Like and subscribe, watch our new video every Tuesday. First of all, let me give an overview regarding AI product I have launched in the past. I launched a product to use AI to reduce car crashes and for Smart Cities application which was awarded as a mayor's best practice award in 2017. And this product was also quickly deployed in more than 10 different cities within half a year and became one of the most famous product using AI to reduce car crashes and also saving thousands of people's lives and maybe really really proud of how we actually apply AI in real life. The specific AI technology we use is called machine vision which is use the smart camera and capture real-time video footage and automatically detect cars, pedestrians and bikers and different object inside of intersections and give awards and analytics for cities to make changes in real-time and making policy changes later on so that they are able to reduce car crashes as a whole. So this day-in-life as AI product manager is assuming going backwards into 2017 when we're still building the product and what does this look like in a day-in-life as a product manager? Now, let's get started with 7am. At 7am, that's when I naturally wake up because my son is 2 years old right now. He will just wake up at 7am and wakes me up and I start to feed my son, have a bonding time with him. Actually, this is the happiest time ever during the day and how I start my day. I will skip workout because my morning is very busy with my son but I will definitely drive my son to daycare, dropping off and giving him a big hug. And my favorite morning drink is actually the cold pressed celery juice out of raw organic celery juice. It has so many nutrition, electrolyte and clean my whole body. I recommend everybody to check it out. Then I get ready to start my work at 9am and my daily stand-up with my engineering team is usually around 9.30am and at the time I have three engineering teams that I need to interact with when I launch this AI product. And the first team is the influencing a product use cases team because this is my engineering team that's more focused on how can they apply the AI which is machine vision in solving the problem of reducing car crashes and they specialize in figuring out how we take real-time video footage and directly use the trained AI model to detect the real-life use cases such as jaywalking, cars running red light, near misses. Majority of time whenever on a daily stand-up is with this specific influencing team. The second engineering team I interface with is AI engineering team and they're specialized in building the AI model and training those AI models before we apply it in the real-life scenario. We don't do daily stand-up but we have weekly sync up I'm gonna show you later regarding how do I interface with them later this afternoon. The third engineering team I interface with is the camera team because all our AI algorithm is running inside of a camera that was able to detect real-life movement of the streets and cars and pedestrians. In this case, they are the hardware team also has embedded software running inside of the hardware and I don't do daily stand-up with them either because hardware takes a longer time to develop and there's a separate hardware product manager and manage those hardware teams independently. During the daily stand-up, my AI influencing team is going to follow the traditional sprint methodology and give me daily updates, how many user stories are working on any kind of roadblocks and this is the same as traditional PM however what's after is very different from traditional PM at 10 30 a.m. I'm going to have a meeting with AI engineering team to check in on the detection algorithm and also the accuracy. Now let me give a real-life demo and show you what is AI machine vision look like. There are some confidential information for the exact AI product launched in the past so I pull some public available images which demonstrates a very similar methodology how we detect those car crashes and for example and this is what an AI inside a smart camera is looking is able to turn the real-life image into pixels that AI and also computer vision is able to recognize such as pedestrians and cars and track movement of where they were in the past moving into a new direction and then once the camera is able to see the real world it's going to start getting trained using our data to say well a picture that look like this is a car a picture looks like this is another car a picture, an image like this very tall is a human this is also human on the sidewalk it can also detect like trees and traffic lights and different trucks different type of vehicles as well for example this is a truck the second thing what AI machine vision need to do is to understand the accuracy of detection you can see many different numbers right here 0.9, 0.42 this means the confidence of what they think this is a car the AI think 91% of the chance this is a car 0.87% of chance this is a human and this is a detection accuracy that we need to continue to train our AI on that's the purpose of me meeting the engineering team to see how fast they have been training the AI and how accurate those detection algorithms are and then they can also report back to me regarding different kind of challenges their facing could be they're running out of data set to train the AI model could be also the weather that leading to the low accuracy detection and all of those I am part of the conversation with our engineering teams and help them to solve problems and make decisions this makes AI partner management so much more fun than traditional partner managers and then the engineering team will also show me how they trained AI model in a sandbox environment because the model is as good as it is it must be test out but you can't test out on the street right away to see how cars run into pedestrian that's a very bad so we build a sandbox environment we start to take the parking lot video image feeding into our AI algorithm and see if can detect cars, the pedestrians correctly and also see the speed of cars and different kind of interactions between cars and pedestrians when they show me the sandbox environment and doing a demo sometime they're going to show me a bunch of code in front of the computers that made me very dizzy because I don't know how to code now it's time to take a lunch break during my lunchtime I love to make my favorite Chinese noodle soup with cilantro and dry grandma hassas that's just dream come true but I always eat very fast and sometime there are some meetings during lunchtime for me to attend and listen in another favorite part on my lunch routine is actually check out my son's messages in the app that we use called Bridewell to communicate with the daycare which they always show really cute pictures of my son so I found really happy in the middle of the day have a little bit treat seeing my son's pictures now it is 1 p.m. it's time to do learning and research as AI product manager because not only we try to just write requirement we also need to help engineering team to solve problems and really think about what are the external vendors and resources we can bring in for example one of the challenges my engineering team told me is that it will take them forever to train the foundation model so they would rather to have some pre-trained model so that they can modify some pre-trained model to develop their AI algorithms on top of much faster and we also need to find specific chips that can run inside of the smart camera that can power very complex AI algorithms so therefore I started to research on NVIDIA and actually we end up partnering with NVIDIA to build application on top of the AI engine now let's take a look at NVIDIA website when we use NVIDIA's technology in 2017 when I built my very first AI smart cities product NVIDIA has a very different kind of product offerings and right now like five years later they already improved their own foundation AI model as well but when you go to NVIDIA website like this they have many different platforms and the solutions they already built AI training, data analytics, inferencing this is amazing what they have right now different kind of AI model services so basically as AI product manager I need to look up regarding different vendors different solutions based on the challenges my engineering team is facing and then reach out to the AI team within NVIDIA such as you click get started and learn more, dive deeper and really do some experiment and demo to see if any of the foundation models they provided any solutions they had can actually solve our problem or not so this requires lots of research and learning of AI knowledge and also understand different kind of competitors in the market right now you may ask me hey Nianxi, Sunxi already know AI very well that's why you can become AI product manager the answer is no actually I am learning lots of AI knowledge on the job and before I became AI PM I also took many different AI classes technical classes out there because I believe that to become a great product manager it's very important to invest in ourselves invest into learning because for learning you can always generate 10 times return of the tuition you put into learning so how much do I really know about AI? I know enough to be vendors but in reality I don't know how to code and my PhD is in material science and engineering I literally start from scratch and that's why I recommend many of you guys take any kind of AI classes even before you become a product manager in AI space so that you have a great confidence to build amazing product to change people's lives using AI and in this video I filmed last time where I talk about the only four way to become an AI product manager and over there I talk many more different kind of AI courses and skills developed before you even become a product manager and also going to link in the description regarding different kind of AI courses I recommend and different kind of technology courses in general that you can download to take on your own as well and those courses really design help you to know enough to become a venture check the link in the description you'll find out more of the courses now after two hours research and technical learning problem solving it's time for me to write some product management requirement and now it's around 3 p.m. I will just go into JIRA and start writing lots of product management requirement and I still recommend everybody to learn JIRA and this is the basic and foundation of everybody being able to write requirement put into JIRA and recently I also discovered another AI tool that's so amazing was able to automate the requirement writing process as a product manager using AI and it's called product monkey so let's take a look how it works so this is product monkey's website the way it works is if you drag any pictures it's able to generate requirement use stories for you so within 30 seconds it generated several two, actually two user stories for me and have user stories, descriptions acceptance criteria and giving another one of course this is based on Twitter assuming I'm trying to build Twitter and then upload image of Twitter and wrote some requirement for you automatically however the challenge I see right now is it's only giving of course two user stories for Twitter you should have hundred of user stories you can write a book about it seriously because of Twitter very complex software environment but this is good enough for inspirations for product managers or even entry-level product manager take a look regarding oh maybe I missed some kind of product requirement and so they can speed up your thinking process I really love those kind of innovation in the AI space creating tools to help product managers to save time and do the job much better I'm gonna put the link of the product in the description you can check it out yourself of course this is the very beginning of testing the product and there's many rooms for improvement but it's great to get started on the AI trend right now now it's 5 p.m. it's my family time my husband will pick up my son from fake care I spent lots of time with him feed him put him to bed and then I continued to work around 9 p.m. and try to wrap up on emails and other tasks I didn't finish doing the day if you're thinking about how to break into AI product management with step-by-step guide and make sure to check out this video where I talk about the only four ways become AI product manager with no experience and also make sure to check out in the description of this video with different kind of AI courses and different tech courses I recommend for all of you guys to brush up your skills and invest in yourself this is Doc Nia CD from p.m. Accelerator.io I'm gonna see you soon in my next video right here