What is machine learning, and why does it matter? Sang Lee, co-founder, and CEO of Qeexo, Co., answers this question for us in this latest episode. Qeexo is the first company to automate end-to-end machine learning for edge devices.
Sang opens up about his multicultural background – growing up in Australia and then coming back to Korea to attend college in a very elite university before going to the US to get his MBA at Haas and experience the Western culture.
Sang also shares his extensive career in the mobile industry. He joined Samsung right after college as a Product Planner, then transferred to SK Telecom and worked as an OEM Partnership Development, eventually moving to HTC to become Technology Program Manager.
We also hear about his company, Qeexo. Sang tells us its early beginnings, how they scaled the business and explains what machine learning is all about.
Finally, Sang tells us the good and bad in managing a company and the challenges and things he learned as a CEO. We also get a tip on how you can tell that you are hiring the right people for your team!
Episode Quotes:
His decision to get an MBA
“One thing that gave me the thirst to come to the US for MBA was not having that business education background. Going to this department where most people have a business background, and you’re one of the few people with an engineering background, you get to question all the decisions you make. Are those people making those decisions because they got that education from school, or they may have learned something different from school that I never learned? You always have this question about your own decision. And I hated that.
I wanted to step up, and Haas really opened my eyes. I believe that it was worth every penny, for me at least, because I’ve started my own company, and I don’t think I would’ve been able to do it without the Haas experience.”
Realizing what his role is as the CEO of the company
“At first, when I was starting the company, I thought my role was going to be product management and sales and maybe strategy. But when we first started, we already knew what product we wanted to launch, so there was no strategy work that I needed to do. But as we grew, I started to learn about what the role of a CEO needs to be to make this business successful. And the thing that I learned is that I’m not a star talent in all of the roles. I’m not a star talent in sales. I’m not a star talent in strategy. I’m probably not the best star talent in product management either. My role is to make sure that I find the star talent and make sure that they perform at their best and work efficiently and effectively as possible.”
On finding innovation through diversity
“We want diversity. That’s one thing that I picked up from Haas. One of the class professors was talking about where the innovation comes from, or one of the ways to find innovation is when people coming from diverse backgrounds or diverse disciplines intersecting, and during that intersection, you find innovation.
And I thought that was a great way to approach finding innovation because when you’re running the company, you actually have the power to find people coming from diverse disciplines and putting them in the same spot. I won’t be able to, by myself, get experience in different disciplines and find innovation within myself. But I can find people with different disciplines coming together.”
Show Links:
Transcript:
(Transcripts may contain a few typographical errors due to audio quality during the podcast recording.)
Sean: Welcome to the OneHaas alumni podcast. I’m your host, Sean Li. And today we’re joined by my cohost, Chris Kim, and Sang Lee.
Chris: We have today the co-founder and CEO of Qeexo, Sang Lee. He is also a Berkeley Haas alumna, class of 2011. Sang, welcome to the podcast. Great to have you on.
Sang: Thanks, Chris. Thanks, Sean. Thanks for giving me the opportunity. Really appreciate it.
Sean: Sang, you’re calling from Korea today, right? That’s amazing.
Sang: Yeah, it’s a rare opportunity for me to work from Korea. Since everybody’s working from home anyway so might as well take the advantage of this coronavirus situation.
Chris: So, Sang, you’re working from Korea. What’s the best thing working from Korea right now?
Sang: I think the best thing is that my family is still in Korea. Like my parents and sisters are all living in Korea. So, I can drop off my kid at my sister’s place and do my own thing. I moved to US in 2009 when I was joining the Haas MBA program. For me and my wife, we don’t have any other family living in US. It’s not that easy to do a startup and raise a kid and have a family life without anybody helping you. It’s definitely an enjoyable experience.
Sean: How old is your kid?
Sang: She’s seven.
Sean: Chris and I are newbies at this so we might have some things to learn from you. I have an 18-month-old. Chris has a 15-month-old, right?
Chris: Yeah, almost 15.
Sang: Oh, wow. It gets better.
Sean: Okay, that’s good to hear.
Chris: So, Sang, you and I met through Haas but I never actually got to ask you this question. When you were growing up in Korea and even going to school in Korea, like, did you know you want to come to the States and be a CEO? Or how did that happen?
Sang: Yeah, I was born in Korea and I lived in Korea until I was 10. And when I was turning 11, our family moved to Australia. So, I also grew up in Australia, went to high school there. I graduated high school in Australia. Then I came back to Korea for college, and then I stayed in Korea for another seven, eight years before I joined Haas. So, I have a little bit diverse or multicultural background as I say. Coming from that Australian experience, I wanted to experience the US. It was me wanting to get a business education and also experiencing this US life. When you’re coming from Asia, obviously, you know, they have a lot of Western culture has a lot of influence in Korea or in Asia.
So, you know, you want to experience that, but like coming from Australia, some of my Korean friends think that, Hey, you already lived in Western culture, what would you want to experience that again? But living in Australia which is like down under, like a Puget Island, I never actually felt that I had that kind of Western, what Asian culture believes as a Western culture.
You know, America and US, this whole thing about American dream and all the high tech Silicon Valley stuff. In Australia, you feel like detached from that so I wanted to come and experience that. I think a lot of it is, things to do with getting the business education and then that cultural experience, and then, you know, going into this entrepreneurship route, joining Samsung right off to college.
I started as a mobile handset product planner and during that timeframe, mobile handset product planning had a lot to do with working with telecom operators, getting the requirements from telecom operated, it’s like Verizon, Sprint, and building things for them.
They were the ones that was actually selling the phones to the customers. It wasn’t Samsung and LG. I mean, I’m sure that other people may have different experience but for me I felt like all the telecom operators has so much power over the device manufacturers that we didn’t really have difficult control over what we were building as are enhancing that we were selling in the market. So, a lot of the requirements I would say probably like 80, 90% of feature requirements for handsets were coming from telecom operators were merely assembling and doing that industrial design and assembling the hardware.
So, naturally I wanted to be on that other side of the table. I want it to be, you know, the product managers at telecom operator giving the requirements to the OEMs. So, after Samsung, I joined SK telecom and was doing that role. And then I saw that, okay, the dynamics in the mobile industry was changing.
That was around 2006, and 2007, we started to see new things coming out, like Apple announcing iPhone. And we were hearing rumors that Google was preparing something and then HTC came out with the, you know, first Android. And then we saw, okay, things are going to change.
And I mean, from my narrow vision, I saw that, okay, within this so-called mobile industry, there is device manufacturers there’s telecom operator. Now we’re going to see the contents. So, at first, my thought was okay, I’ve experienced two of the three within this mobile industry. I want to go to the US and get a proper education on the business management, and then, you know, start my own thing, hopefully in the mobile industry. That was my plan. And that’s how I came about to join Haas and MBA.
Chris: Pretty cool.
Sean: That was actually really interesting the whole time I was thinking, what year was this? I mean, it must have been before smartphones. And I was curious how things actually have changed after the smartphones came out after Samsung becomes such a big force in mobile technology. Is it still dictated by telecoms?
Sang: So, when I first joined Samsung, it was 2003. I still remember when I first joined the team, the head of product planning was saying, oh, last year we sold 27 million units. This year, we’re going to sell 54 million units. And I was thinking, this is crazy.
Like, how do you double the sales volume? And then they were doubling, doubling, doubling. And it was going up to like, I think 350 million units a year or something.
During 2003 until I think 2007, it was mostly featured phones, folders flips, and all this candy bar design. So, it was all about, you know, designing some fantasy industrial design and then most of the features were, I didn’t really think that there was that much difference in terms of the features among different brands, it was mostly the design. So, the customers were mostly looking at, okay, what is the megapixel of this camera, what’s the actual physical size of the screen, and the resolution of the screen.
That was like the most selling point. And then in 2007, when iPhone 1st was announced, then everybody was like in this panic mode, you know. We were like writing reports trying to figure out, Hey, is this really going to stick, or is this just like one-time thing? And at that timeframe, I already switched my job to SK Telekom.
And it was easier on SK telecom side because as a telecom operator, OEMs are bringing in the reports, they’re saying, oh, we’re going to change our roadmap. What are we going to do this? And do that. And during the timeframe, we were seeing a lot of the OEMs trying to do all of it. Like they were trying to get into Android. They were trying to do Windows and they were also sticking with feature phones. They didn’t want to make a bet on just one single thing. So, there were a lot of different experiments that was done in the market. And then I think what happened was, obviously, opener Wes platform, took his way and everybody’s switched to Android.
But there were a lot of struggles like Samsung struggled with using Windows for some time. And they were getting criticized a lot, I mean, it wasn’t a fair game for them. They were basically trying to launch something that looked like iPhone but had windows and it wasn’t working well.
And market didn’t really like it but eventually, they were the right ones. I mean, if there was anybody that can do anything close to the Apple, it was going to be a Samsung. So, they had engineering, the skillsets, and the team, and the experience.
And so, when Android eventually became mature, they were the ones that was able to take advantage of it. But I think after the first couple of years of this old trials, lot of this happened while I was in MBA. So, when I came to the US in 2009 until 2011, the market was changing a lot.
And then 2011, I joined HTC. And then I saw that HTC during the idea to launch, even the in 2011, beginning of 2011, HTC was still seen as one of the top powerhouses in Android device. But at the end of 2011, when I joined them, they had many different models that had, all had great hardware, great OS. But Samsung also had the same OS. Samsung also had same hardware. LG had the same thing, like China, a lot of new brands were coming out with the same thing.
Only difference there was Samsung knew how to do the marketing. 2011, I think it was the first year when Samsung started to use the brand name Galaxy. So, I was at HTC and I saw that, Hey, we have a lot of great models, but at the end of the year, when you think about, okay, which model did HTC launched during that year? Nobody could remember any name of the model because there were so many different models with so many different designs. And, you know, there wasn’t much emphasis on the branding whereas Apple, there’s iPhone, Samsung does Galaxy.
And then I think it just changed the whole game. So, I think overall, it was a very exciting period for people that were in the industry. Also, it was a very painful period. But during that timeframe, like 2006 until 2000, I think 11, 12, there was so much uncertainty and things still have changed a lot since then.
You know, after 2011, 2012, the new brands that came out from China also took over a lot of the market share and today, a lot of things have changed.
Chris: That’s crazy. I know we didn’t really talk about it Sang but for folks who are familiar with Korea, you actually went to a very, very competitive, very elite university in Korea. What was that like coming from Australia to Korea? And then you were in college, did you know you wanted to be in mobile devices?
Did you know you wanted to be in like cellphones or was it just like, Hey, I’m studying engineering and that’s like the natural path? Where are you thinking, Oh, I want to go to Samsung right after college? You went from Postech and then you went to Samsung. I feel like that’s a dream for a lot of young people, growing up in Korea. That is.
Sang: Yeah. It’s definitely a great question. I definitely need to open this dark side of my background.
So, when our family moved to Australia, I think you guys coming from your ethnic background, you probably know that like the kids that grew up in Asia, during the early school years, they go through a more intense education in terms of like math and science since I on. So, I finished my fifth grade in Korea and then moved to Australia and I joined the elementary school in the sixth grade, like the math and science, I was in a different level. I was doing like calculus.
Sean: Okay.
Sang: Because I was also going through a lot of like private touring, like my mom’s like all fanatic about like education. So, in Australia, I always saw myself as the top-of-the-class kid. If my grades dropped, it wouldn’t be my standard, I would think that, okay, I need to study and do harder cause I cannot be dropping on the ranks.
Sean: Yeah.
Sang: So, I had that mentality. But when I joined Postech, Postech is much like Caltech, benchmarked of Caltech when they first built a school.
So, there’s only 10 departments, all engineering and natural science departments. And everybody who’s in the school are math genius and physics genius. So, first exam, I still remember I went to electrical engineering department and one of my friends who is now a dentist, we’re taking this math class, and the average score for that first exam was like 30. And my score were 16. My buddy had 96 and he was complaining that he didn’t get a hundred. I was like, oh my God, this is crazy. At that point, like I was hanging out with my friends fairly well and making friends there. But I saw that, okay, these guys, they are in a different level. Like, I thought I was the genius one but it was a completely different, different league.
That’s when I thought, okay, maybe I’m not fit for engineering. I need to go into something else. I actually enjoy school a lot but then when I was graduating Postech, most of the friends there went on to graduate school.
I was the rare ones that was just going to graduate with the bachelor’s degree and joining company. So, I’m one of those like ugly ducklings. They’ll think of it as, oh man, that guy didn’t go to graduate school. So, when I joined Samsung, wanted to move on to something different.
And then I was lucky enough to meet a great HR manager there who saw that I had different interests other than engineering. So, even though they recruited me as an engineer, he gave me a chance to join product planning because product planning is one of those positions where having business acumen and also engineering background is really much needed.
Sean: Wow.
Chris: That’s awesome. When you were young, did that push you to come to the States as well? Cause I feel like that’s also a huge risk to kind of worked at two of the biggest. SK telecom is probably one of the biggest telecom companies in Korea. Why risk all of that to come to the states? I feel like people would have been like, Hey, that’s a huge risk. You’re flying high and you’re doing well.
Sang: Yeah, you’re right. I definitely gained a lot working at Samsung. I learned a lot about how to set your mindset as a product planner, what it takes to be a product planner, understanding the market needs and really building it into the requirements and communicating with the engineers, and launching the product.
So, we all thought of the product as our baby, especially when you’re working on a product that you will see in the market, seeing your friends and family using it every day, you do pay a lot of attention to what you build. But one thing that gave me the thirst to come to the US for MBA was not having that business education background. You, going to this department where everybody like, well, most of the people have business background and, you know, you’re the few number of people that has engineering background.
And then you get to question all the decisions that you make. Are those people making those decisions because they got that education from school or they may have learned something different from school that I never learned? And you always have this question about your own decision.
And, you know, I hated that. And I hated asking them like, Hey, did you learn this from school? And everybody’s saying, no, no, no, I didn’t learn it from school. And, but at the time, because I didn’t get that education, I feel like no, that guy must’ve learned that from some professor. And so, even when I moved to SK Telecom, I still had that eagerness to come and get this education.
So, SK Telecom life was great. I sometimes do regret leaving SK Telecom. It was like awesome but it was the right time for me. I wanted to step up and Haas really opened my eyes. I do believe that yeah, it was worth every penny for me at least because I’ve started my own company and I don’t think I would’ve been able to do it without the Haas experience.
Sean: By the way, it’s funny that you had that experience where you were an engineer and you felt like you had that deficiency in business. I’m kind of the other way around, I have that finance and business degree and I’m working on a tech startup now. I’m always wondering, well, I wish I knew more engineering psych and think like an engineer and like problems like an engineer and really wish I could go get an engineering degree sometimes.
Sang: Yeah. I mean, I think it’s building confidence through experience. I think that’s very important cause once you’re experienced you, you know that, oh, it wasn’t really that big of a deal. And it’s not that difficult to work with an engineer even if you don’t have engineering background. But it’s overcoming that fear I think that requires experience. Okay.
Chris: Same. When you were thinking about business school, I think a lot of people coming from Korea typically go to the east coast. Why did you end up coming to Haas? Did you know you wanted to be in Silicon Valley? Was that the big draw or was it at all the options, it seemed like the best option for you. How did you figure that out?
Sang: Yes, it’s Silicon Valley was the big thing. And also, I had a friend from Samsung, former colleague that went to Haas. He was class of 2008. He was giving me a lot of advice and he joined Google in 2010 and it was telling me about all this like Silicon Valley story and the lifestyle and I think one thing that really got me was that he was doing some blogs at that time like he was posting blogs every now and then. And one day, like he was posting blog of picture of like Steve Jobs sitting somewhere like eating lunch. And then like the other day is like guy Kawasaki, like ordering something from a yogurt place.
And he was like, oh, this is like every day, it’s like a valley. I was like, oh my God, I need to go there right now.
Chris: So, your heart was set on the bay area, like from the get-go.
Sang: Yes.
Chris: You’re dreaming that up while you were pulling hours in Korea and then finally getting over here.
Sang: Yeah. It took me about a year and a half to prepare. Yeah, it was great. Once I got here like it was all I imagined.
Sean: I think it’s a good time to pivot our conversation a little bit into what your company’s about. What Qeexo is about? How did you get into this? I mean, it seems like a very natural progression from hardware. Like you mentioned earlier, mobile hardware to the telecom now to software, almost of sorts. How did you get started?
Sang: So, today at Qeexo, we do machine learning platform service and also machine learning application business. When we first started Qeexo, the whole concept or the whole business model that we started with was a little bit different from what we do work today. When I graduated from Haas, I wanted to start my own company but I didn’t have the right setup. I didn’t have the right idea or the right business that I wanted to do. I wasn’t prepared to launch something right after school. That’s why I joined HTC.
I was doing technology sourcing, they called it a technology program management. There are managers that are going out there’s, you know, scouting and sourcing new technologies. And I met my co-founder while looking for new solutions in the mobile handset, user interaction. And my co-founder Chris Harrison back then was doing Ph.D. program at Carnegie Mellon University. And he had great ideas. We met and, you know, it was just cold email. I emailed him and told him that, Hey, this is Sang, I work for HTC. I’m looking for cool technology. You seem like you’ve got a good technology, let’s chat. And coincidentally he was in Seattle cause he was visiting Microsoft.
So, we just met and had coffee and we were just talking about the cool things that’s happening in the world. And then I guess, when you find someone that’s going to be your co-founder, then, you know, something clicks. The conversation is just going smoothly.
And the whole philosophy about coming out with innovation was matching. And we were talking about things in the mobile space and also in AI and, you know, everything just clicked. So, after a while, was just keeping dialogue with him over email for a couple of months.
And then, I told Chris that, you know, one of the technologies that you have is something that I think I can commercialize in the mobile space. So, let’s start up a company. So, that was a technology that we launched in 2015 with Huawei. It’s called FingerSense.
So, back then it was a research project that Chris did at Carnegie Mellon. He wanted to figure out a way to add additional input dimension or support additional input dimension on the touchscreen device using vibration signature. So, acoustic signature. So, he had that whole experiment done and he wrote a paper about it.
And when I read it, I thought, okay, this is a great product that can add huge value to a mobile device. So, even though his original research was mostly done on much bigger screen, like, interactive surface type of screen, I thought that that could be a great fit for a mobile device. So, that’s how we started the company.
That was the first product that we built the business model around. Basically, it took us about two and a half years to commercialize the product. And we launched a product with Huawei. Huawei has been using our fingers and solution since 2015, even until today is one of the top sales feature on Huawei phones. I think we sold little over 400 million units of mobile devices that are using our technology.
But then in 2018, we started to prepare to do a pivot. So, FingerSense the application business, it’s IP licensing business model, and it’s been doing well but what we saw in the market was a bigger opportunity than just the application.
So, when you look at AI and machine learning, Qeexo solution, the first application, the FingerSense, the users, they don’t need to know what’s happening behind the scene.
All they see is that when you touch the screen with your fingertip or knuckle, it understands the different part of your finger as a different type of input. So, kind of like having left click and right click on a mouse. So, without having an accessory, you can basically tap on the screen with your fingertip and the screen will immediately know that, okay, Sang is using fingertip. If I touch it with a knuckle, it will immediately know that, okay, it’s knuckle. So, if I just assign left click on the tip, right click on the knuckle, there’s left click and right click, right? So, the touch screens already understand the XY coordinates of where you’re touching the screen. What the touchscreens do.
But if the touchscreen can tell XY and type, then applications can do more with that type of information. And how do we do it? Very simple. There are lots of sensors in the phone. We don’t make any hardware. There’s this sensor called motion sensor. It’s basically an accelerometer sensor and a gyroscope. You turn on the data stream, and then we ask the users to tap on the screen with your fingertip, we collect data. We ask the users to tap on the screen with the knuckle, collect data, and we’ll build a machine learning classifier that can classify fingertip versus knuckle. And Chris, our CTO, when he first came up with the idea, that’s basically what he did.
He got a big glass table, stuck a stethoscope under the table to collect sound data, and then he started to tap on the glass with like ping-pong ball, brush, finger, and hammer, and all sorts of stuff and wanted to test whether a machine learning classifier can classify the different type of input and he was able to do that.
And that was his idea. And then from there, I thought, okay, Hey, let’s see if we can do fingertip and knuckle. And if we can do this on a mobile phone, then OEMs will love it. So, that’s the whole background technology of FingerSense application.
But the difficult part about doing a business with machine learning technology is not about building the solution, it’s all about scaling, that we learned that hard way is when we first went to Huawei, we told Huawei, look at this, we built a demo. It’s working well, Hey, you should use it. They’re like, okay, we want to use it, come and do the implementation. And we went there and then we realized that, okay, let’s implement this. And our implementation process requires, I cannot give you the exact number but X number of people to give data on the actual phone.
To build a life. Cause we don’t want people to do calibration after they purchase the phone. We don’t like that experience. I mean, we are all about bringing new better experience. We don’t want people to launch the phone and then they have to do calibration, that’s bad experience already. So, this has to work right up out of the box.
So, we need a lot of people to come in and give data so we can pre-build the library that works well for 99.9% of the people. Now, when we did the first implementation, 2015, we’re seven people team, a small startup. Everybody had to fly to China. We lived there for two months and we were able to optimize a solution for one model.
After finishing the one model, Huawei told us that we need to support 60 models here. And there was no way that we can support 60 models. We just brought entire team to China and we lived here for two months and we only finished one model. At best we can do six models here.
Chris: So, for context, you guys were a small startup and you were just picking up and moving to China, living in China for months at a time, and then coming back, just to kind of relax and recharge, right.
Sang: Exactly. Yes. I mean, we went through a lot of evaluation projects in Asia, in Korea, in China, but when it came to do commercial implementation, the level of quality expectation is really high for a mobile device. You can imagine like there’s hundreds of thousands people using it or millions of people using it.
So, their quality level is probably at an extreme point. So, optimization work that we have to do was very intense. Collecting data, building the library, it doesn’t take that long, maybe at maximum, a week. But then they do it quality test. If it doesn’t work well, we may have to recollect data from scratch, rebuild everything from scratch.
And we were going through that iteration multiple times. And our team wasn’t equipped to do that. Like they were getting burned out pretty quickly. So, after the first model, after doing it two months in test, work, launching it, and we did a grand launch in Europe and, you know, everybody’s like proud and happy and so on.
But then our team was like, are we sure we want to do this forever? Cause it’s, one of the guy was saying, they, I don’t think I can work for kicks at any money to find another job. So, we saw that, Okay, if we want to scale, we need to do this in a more efficient way. We cannot have machine learning engineers to be living in China for two months just to finish one model. We’ll never be able to scale, right?
So, at that point we decided, okay, we are going to have to build automated platform so that our engineers would not have to do this manually. There are a lot of things that still needs to be done manually but also there were a lot of things that can be done in an automated fashion.
So, we worked on building automated platform since that day and we were supporting Huawei along the way and building automated platform. So, that platform that we were building was basically designed for our engineers to use as an internal engineering tool. So, after a couple of years, in 2018, that was the first year that we actually launched 60 models with Huawei.
We were able to launch 60 models without having anybody to travel to China cause everything is already automated. And we have field engineers in China, basically, right after college, non-machine learning engineers that can just go to Huawei, collect sample data, upload it to our platform, and library will be built and sent to them within two hours.
So, this turnaround time was really quick. We were able to build high-quality solution that’s being used in a mobile phone. So, that’s when we saw that, okay, we solve this major problem in AI industry, which is the scalability problem. And we have proven that it works in the market. Maybe this platform is worth more than the application.
Because there are other machine learning companies that’s facing the same problem. They’re going through the same thing. And you can find when you have an idea that requires machine learning engineers to build, I can guarantee you, you can go to any computer science department in Berkeley or Stanford and then find the student and they’ll be able to build it for you because there are tools out there that’s really well-built and easy to use.
Only problem is they can build a solution for one environment. But if you want to apply the same solution in a different environment, they’ll have to go through the same process to build everything from scratch. Collecting new data from the new environment, and building everything from zero. That’s the scalability problem, and that we can solve for the companies.
So, starting from 2019, we decided, okay, let’s repackage our tool and make it into a serviceable platform. So, that’s what we did. And in 2020 June, we commercialized our platform as a service. And now we’re working with customers to allow our customers to use our platform in the field.
Sean: I’m so glad you shared that story. I mean, these are the stories that we want to hear. People can’t see us, obviously, but my jaw was just open. I was like, wow, this is so interesting.
Can you tell us a little bit more about what exactly is the solution? As a company, let’s say, me as a startup, can I use this solution?
Sang: So, we specialized in building tiny ML solution. This tiny ML term is fairly new. So, for those audience that they may not be familiar with tiny ML term, you can look it up. It’s tinyml.org is the place where they started this, the whole tiny ML movement.
So, tiny ML is one of the areas within the whole AI machine learning industry that focus on building machine learning solutions that can fit into a very small microcontroller, or fit into a hardware that has very small computation power. So, there are lots of different things done in AI industry as you know. There’s natural language processing and there’s video processing and all those are called application side of machine learning.
And a lot of the machine learning engineering and processing today for grand scale applications are done in a cloud server with a huge computation power, with lots of graphics cards, and so on. There are very specific needs in the world that require machine learning to run on the edge device. And for those types of applications, you do need to build a machine learning solution that is very, very small so that it can fit into a very affordable hardware. Like for example, when you have a motion sensor like accelerometer sensor or gyroscope, those sensors have a small microcontroller that’s attached to the sensor that drives a sensor.
If you can have an AI solution that can run on that microcontroller, then you don’t need to send the role of sensor data up to the server to process that and run your AI engine. What we do at Qeexo is we build AI solutions, that tiny ML solutions that can run on a small microcontroller. So, our platform is designed to build something that can fit into a very small microcontroller.
So, if you’re an application company or if you have an idea for machine learning application and you want to build this and if it requires sensors to collect data and do some kind of inference on the edge, you can use our platform to select the hardware that you want to use and collect data from field.
And then you can just do the point and click with your mouse button to build your own machine learning library that can run on that hardware. One of the application example would be, something that we’re not doing so you know, something that I can share with everyone, is that I heard from someone that there is a company that is working on doing road condition classifications with a microphone and machine learning solution.
So, let’s say you’re driving Porsche and you have a microphone attached to rim, and you want the traction control to switch automatically depending on the road condition. Whether the road is wet or icy or dry. The way you want to do this is you use a microphone attached to the rim, it will be collecting the road noise, and of course, the wet road versus the icy road sound is probably different. If you can have a machine learning engine to do that classification, then you can just do the classification right on the edge and send it to your thurst and control manager on the, you know, your Porsche, and then it will be able to automatically switch the thrust and control so it’s safer for you.
So, when you want to build something like that, you can use our platform because it allows you to collect the data and build a library that can run on that specific hardware.
Sean: I see. So, how similar or different is this from let’s say Apple iPhone. They talk a lot about their neuro networks or like the camera, how they can composite images from the three cameras on the device. Is that leveraging machine learning as well?
Sang: I don’t exactly know how Apple’s doing it. I can only speak for our technology but an easy way to put it is this – you have five natural senses; if you can tell the difference between one or the other using your natural senses with sensors and machine learning and AI, AI can probably do better, better than human.
Like for example, you can tell the difference between Chris’s voice and my voice. With a microphone and AI, it will be able to tell the difference even better. Cause sometimes for human, consistency can vary when you get tired and when you have fatigue, but robots, machines, they never get tired.
[00:34:00] Chris: Never get tired.
Sean: That’s a great example. And sorry for my naive question, I was just thinking about it but actually after I said it, I realized I don’t think that’s machine learning. That’s just, they have some kind of process in place. Whereas machine learning is it’s learning and adapting continuously.
Sang: Yeah. But I think they may use machine learning in doing that post-processing. There may be some machine learning that’s going into that. I don’t know. But also in machine learning, sometimes it’s not always ongoing learning, like for example, the solution that we sent to or we launched with Huawei is not an ongoing learning process.
It’s a prebuilt, fixed library so that there is no ongoing learning, but yes, lot of the times today, the applications that we are familiar in the AI and machine learning industry, we always learn about how the robot was able to adapt to a different environment or like AlphaGo learning from the previous games that people have played or learning from how people are playing today. So, there could be an ongoing learning process.
Tiny ML space is a little bit challenging to do ongoing learning because everything needs to run on that microcontroller. We’re talking about library size that’s less than 10 kilobytes. So, we’re not talking about Nvidia GPUs or the new M1 processor. We’re talking about small microcontrollers.
They can only do only few types of computation. So, the difficulty here is about how to build a library that is small yet effective in processing and building the application that you would desire to build.
Chris: Sang, I know when we first met, and you explained what your company did, I was telling Sean I was just blown away. We were sitting I think, in your conference room in your office. And I was just like, oh my gosh, you’ve because it’s so hard. The product itself is solving such a hard problem, but also like managing a company, what are some of the inside scoops? What’s the good, what’s the bad, what’s the ugly in terms of what it takes to run a company because I think a lot of people see stuff on TV or they hear about the person who started a company and sold their company in six months, something like that.
You’ve been a CEO and founder for a while now. So, what’s the story about running your own company and running a company with a lot of other people as well.
Sang: We have many different office locations. We have office in Mountain View, Pittsburgh, Shanghai, and Beijing. And also, we have engineering career too. So, I always tell people who’s asking about entrepreneurship experience that my story is just my story. Everybody else has different stories. So, I can only speak for myself.
But I had to operate in this environment where people are scattered around. So, we weren’t able to really put everybody in this same spot and work together as a team in the bay area. The difficulty that I had is because of this environment, I learned that you know, a lot of the things that I have to do as a CEO was about making the operation go as smooth as possible. How the team can work together, even when they’re in a remote space, remote location, and making sure that I’m providing all the resources that our team needs in order for them to perform at their best.
So, at first, you know, when I was starting the company, I thought my role was going to be product management and sales and maybe strategy. But when we first started, we already know what product that we wanted to launch. So, there was no strategy work that I needed to do. I just needed to make sure that our engineers are building the things that we can launch.
But as time went by, as we grew, I started to learn about what the role of a CEO needs to be in order to make this business successful. And the thing that I learned is that I’m not a star talent in all of the roles. I’m not a star talent in sales. I’m not a star talent in strategy. I’m probably not the best star talent in the product management either. My role is to make sure that I find the star talent and making sure that they perform at their best and they work efficiently and effectively as possible.
So, suddenly, I found myself being the operation guy, office manager. If there’s anything that breaks, then I’m the one to go and fix it. I don’t want anyone else to move the furniture or fixing stuff because that’s wasting their time. Our engineers are at best building machine learning solutions.
So, I learned that along the way. And I guess I will say the valuable experience and some hard experience that I had to go through to find out, Hey, what is really my role? And I guess in the beginning that was the right role. In the beginning, I had to dump out my assets, my network, and my sales skills and strategy in the mobile space in that was a way to get us up and running. But once they started to run then I need to figure out how are we going to scale and grow as a team?
And if I want it to be the key talent in the sales, then I don’t have time to manage the team. So, finding out what my real role is at the company, that was the hard part.
Sean: Do you have any tips or advice on best practices for hiring a star talent?
Sang: Yeah. So, we have one philosophy in Qeexo and I picked this up, I don’t know when but I picked this up a long time ago from somebody. It was so long ago that I forgot who it was but basically when we are hiring, we always tell our team to ask themselves this question when they’re interviewing – is this the person that I want to invite to my barbecue party this weekend?
And it doesn’t mean that they need to invite them for a barbecue party, in fact, I’d never had a barbecue party myself. Engineering skillsets, we go through all these rigorous interviews with coding skills and so on. And for business, there are a set of questions that you can ask, and honestly, I think if people practice, they’ll be able to answer most of the questions with the perfect answer. It’s not like there’s a unique question just for Qeexo. I want to make sure that when we are hiring, we’re hiring the team that we’re comfortable working with.
And we want to work with somebody that’s pleasant to work with sitting next to, spending the time together, having good conversation. And if you think about it, like during the short time during the interview, that’s very difficult to pick up. Sometimes I feel like after five minutes you either know or you’re like, oh my God, I don’t think I really know this person well. But if you think about, Hey, is this a person I’ll be comfortable like inviting over for barbecue party and I would have no problem that guy standing on the corner and wouldn’t feel like that would be awkward? Then maybe this is the guy.
Sang: Oh, and one more thing I should mention is that we want diversity. That’s one thing that I picked up from Haas. One of the class professors was talking about where the innovation comes or one of the ways to find innovation is when people coming from diverse backgrounds or diverse disciplines intersecting, during that intersection, know, you find innovation.
And I thought that was a great way to approach finding innovation. Because when you’re running the company, you actually have the power to find people with coming from diverse disciplines and putting them in the same spot. I won’t be able to, by myself, get experience in different disciplines and find innovation within myself. But I can find people with different disciplines coming together. And that’s actually one of the things that, Chris, our co-founder, and I talked about when we were incorporating the company. And because of that, we were able to hire a lot of good machine learning engineers that are coming from different disciplines.
So, not necessarily machine learning engineers with a computer science background but we were looking for those machine learning engineers that’s coming from different discipline but have been using machine learning as a main tool for their research. One of our engineers in Pittsburgh has Ph.D. in medical science, another one from physics, another one from economics.
But they all use machine learning as the main tool for their research. So, when we’re all sitting in the same room talking about machine learning, they have been using machine learning in a different way. So, we can find new things, I mean, it’s shame that we weren’t able to have face-to-face conversation for like a year and a half. But every time I’m in Pittsburgh there’s always a new topic, some random topic because everybody’s looking at it from different magazines and their sources are all different. I think that’s where the innovation really comes.
Sean: I love that. We always hear these days the benefits of diversity in your teams and your organization. But I think just the way the professor phrased it and framed it, I thought very unique as this is actually something that is extremely beneficial for innovation. Not just for your company as a whole but for progress. Do you have any other questions, Chris?
Chris: I don’t know. I’m just so motivated right now.
Sean: Same here.
Chris: Sang, one of the things I always appreciate it and we’ve only been able to connect a couple of times since the pandemic, like right before, and I think couple of weeks ago we connected. Yeah, I feel like that passion and that drive, feel it must be a pre-req to be a CEO or co-founder.
Cause I know Sean, you definitely have that internal fire going on. And Sang, I feel like every time I talk to you about whatever it is, like, just have so much fire. I mean, feel like that’s a trait that all CEOs or all founders have to have cause you just have so much conviction from what you’re doing.
Sang: Thank you for the feedback but to be very frank, I think the CEOs I think do seem like more energetic and more passionate because we do get to talk about the same thing over and over again, many, many times. I mean, I don’t need a script to talk about my company.
I didn’t need a script to talk about myself and I can tell you, in the beginning, it was very awkward. And sometimes, people watching me going through the interview was telling me, Hey, Sang, you need more energy. Cause I have to tell the story out of the script and I didn’t know what were the things that I can say and not. The pitch that I make, you know, I’ve been making the same pitch to different investors, hundreds of times. As time goes by, you know, it just becomes natural. Because I already know the content so well that I can just burst it out and then feel like, oh man, this guy has energy, but it’s not always what it looks like.
Sean: I think Sang is absolutely right. You listened to yourself talk enough times the same story you can’t help believe in it. Right? It’s like it is here over and over and over again. That’s a great point.
Sang: Yeah, over the time, you know that the solution that you built, you have thought about the problem many, many times and the people that are listening to it for a lot of them it’s first time they ever heard about this. So, well it seemed like the CEOs are more excited but I also tell the team that in order to motivate and give a morale boost to the team, regardless of what you really believe inside. Like I told the team, Hey, I believe in this technology and the team 200%, there’s no doubt. Inside. Like, I may think that I’m like, ah, what am I going to do for next month? And there’s all sorts of trouble than issues that I have to think about. But that’s the hard part, which I didn’t mention when you asked me about the hard part of running the company, but the really the hard part is something that I cannot talk about.
And I actually listened to one of the podcasts today, one of the entrepreneurs was talking about they want to bring, just need to get a therapist. Lot of people told me this before, and I didn’t get a therapist, but I feel like even getting the therapist, the thought about it gives me a fright. That’s the amount of pressure and stress that I know that I need to get a therapist but getting there already spooks me. Confronting myself, saying the things that’s inside, oh, man. I don’t think I can handle that.
Sean: That’s very true. And I think it is true that everyone has to take care of themselves. Speaking of which, I mean, last month was mental health awareness month. And just on that same topic, been thinking about this for a while. And I wonder if it’s a cultural thing. I’m from China. You guys are from Korea. But I grew up here in the US and there definitely is a lot of stigma around mental health, especially in Western culture that I found because it’s such an individualistic society.
You’re supposed to be individual and strong versus at least in Asian communities, there is more a community sense. But then what you’re dealing with there is shame. You don’t want to shame your family by appearing too weak. But think we’re coming to a new age of sorts where we recognize that for our employees, for our companies, for our families, to be healthy, we have to be.
Sang: Yeah, I agree. I definitely agree. And I definitely need some help, I need some advice from you guys. I feel like maybe it’s just probably my unique position here at Qeexo. So, but I feel like a lot of the CEOs would probably feel that there’s always this thing that you’re not able to share with even the co-founder because of the CEO role. And those are the times where you feel like it could be a lonely cause there’s board members, investors, and then there’s the team. And you’re like in the spot and there’s nothing that you can do about this. Like you have to figure out how you’re going to communicate with board members, investors, and the team.
It’s not like you can tell the board members, Hey, how do you want me to address this? So, those kinds of things. It’s not like these problems come up every day, but it is there.
Sean: I think it’s great that you have self-awareness so that’s a great skill.
Chris: I feel like one of the things that I’ve taken out of the pandemic especially, is the power of friendship. Especially for like therapists and therapy, it’s great to have someone who can sit down with you but in the moments where you can’t have that interaction, the power of friendship and community, I think Sean, you’ve talked about this a ton of times in the past, but just like that power of having people who care for you, who don’t need to get anything from you but you could go to with anything.
I mean, I’ve definitely experienced the positive benefits of it more than a year now that we’ve been in this situation. But yeah, I think it’s just been a good reminder that even though it may not always seem like there are people out there who really care for you in that way and who can really care for you unconditionally, there probably is at least one or couple of people out there and we shouldn’t take them for granted, but at the same time, we don’t have to always do it alone because there is probably someone in our community and whether it’s a Haasie for us or someone outside who we could go to and just share candidly and at least know that someone is there for us.
Sean: Yeah.
Sang: Yeah, I agree. I definitely had a lot of help, old relief, from talking to other entrepreneurs. My friends in the bay area who are also doing a startup, it’s easier for them to share those stories and letting them know that, Hey, our investor just told me this or one of my managers just left.
And then, you know, they’ll say their story and there’s definitely a feeling of more understanding when you have somebody that’s going through the similar thing.
Sean: That’s great. Well, my last question to end the podcast on a light note is how did you come up with the name Qeexo and what does it mean?
Sang: Our CTO Chris came up with the name. The word qeexi means analysis in Somali. In the beginning, we want it to come up with the name that represents the work that we do and also the adventurous culture that we want to establish in Qeexo.
Chris has a lot of different types of experience. His parents are professors at NYU. And when he was young, his parents were taking him to these expeditions in Africa to dig up like human bones and so on. So, he had a lot of experience. His dad is literally the living Indiana Jones.
So, he had an experience in Africa. So, he came up with the worst thing. Hey, in Africa, there’s this sporcle qeexi, it says analysis. Let’s name our company Qeexo.
Sean: Love it.
Chris: Awesome. Hey, Sang, want to thank you again for joining us on the podcast and wish you and the rest of the Qeexo team all the best. Go bears.
Sean: Go bears.
Sang: Chris. Thanks, Sean. Really appreciate it.