Leading AI Innovation from Intel to Nota America with Steven Kim

Sheikh Shuvo:
Hi, everyone. Welcome to another episode of Humans of AI. I'm Sheikh Shuvo and we'll be talking to the leaders who are building the technology that's changing our world today. I'm pleased to share Steven Kim with you, the CEO of Nota America. Nota is a leader in hardware-aware AI optimization and edge solutions.

Steven, thanks for taking the time to join us.

Steven Kim:
Thank you for inviting me.

Sheikh Shuvo:
Can you hear me okay? Yeah, there was just a brief pause, but we're good to go. Okay. Well, Steven, the very first question I have for you is, obviously Nota does so many different things. If you had to describe your job to a 5-year-old, what would you say you do?

Steven Kim:
I do get this question a lot from my kids. So, what do you do? What does your company do? The way I explain to my kids is that we take really large AI models and we compress them in a way that consumes less energy but performs better. So it can be used in a number of different places, including edge computing. We're talking about these things.

Sheikh Shuvo:
Nice. And how old are your kids?

Steven Kim:
Seven and nine. They don't understand the concept of energy computing, but what I tell them is our models go into traffic cameras, facial recognition on their door locks, and driver monitoring systems for cars like Teslas, and things like that. Those types of examples help them understand what we do.

Sheikh Shuvo:
Yeah, I'm just a couple of years behind you. My girls are six and eight. I'll have to come to you for tips on how to talk to them about tech. Of course. Sure. Well, cool. Tell us about your career story and how you landed where you are. It must've been quite a journey.

Steven Kim:
Yeah. So, I was a physics major in undergrad and I got my engineering degree from the University of Michigan. My first job was at Intel in 2000, working as a process engineer in the fab. Then I got my business degree from Columbia during the economic recession in 2008. I ended up going back to Korea to work for Samsung and worked as a corporate development team member at Samsung Electronics. Then, in 2016, I took a leap of faith and started working in the startup world. This company, Nota AI, my co-founders were looking for someone with more experience than they had. I wanted to bring my experience to the table so that the company has a really solid direction in terms of where it needs to go and grow.

Sheikh Shuvo:
Nice, nice. Awesome. Now, you mentioned you were at the University of Michigan. What was your experience there? I actually used to live in Detroit and Ann Arbor for many years. It's always fun to meet another person who knows Michigan.

Steven Kim:
Yeah, I spent my graduate years at the University of Michigan. I studied material science and concurrent engineering or radiological science. I was a dual major in Michigan, so I spent most of my time on the North campus, working as a research assistant and as a student in these two different engineering departments.

Sheikh Shuvo:
Nice. I feel more than any other school in the world, whenever I travel around the world, I see the most University of Michigan t-shirts.

Steven Kim:
Yeah, I love their logos and I used to go to their games a lot. I'm a huge fan of the University of Michigan football team and their athletic teams.

Sheikh Shuvo:
In working at Intel and then at Samsung, when you started diving into the startup world, what are some of the differences that you noticed there? Were there any major changes in how you worked going from a corporate to a startup setting?

Steven Kim:
You know, I knew I was getting into something that I had no experience in, but I didn't realize how much I didn't know. In terms of the types of challenges I had to face, when you join a startup, everything has to change about you, except for your family members and things like that. I had to change basically everything about myself in terms of how I work, how I talk to people. In a corporate environment, there are really strict rules about how to make things work. At startups, there are no rules. Whatever works, you have to apply them and learn everything on the job because no one teaches you anything. It's like jumping out of an airplane and before you hit the ground, you have to assemble a parachute so you don't actually die. If you don't do that within a given time, your company will not be successful. So, I'm always under this pressure to produce results within a given time with very little resources. I had to do everything, which I didn't have to do in a corporate environment. I realized how sheltered my life was for the first 16 years of my career until I joined a startup.

Sheikh Shuvo:
Well, it sounds like your parachute works at least, and you're floating above many times.

Steven Kim:
I crash. I learned from a couple of mistakes that I made. So, when I joined Nota, I tried not to repeat the kind of mistakes that I made earlier.

Sheikh Shuvo:
Yeah. Well, looking at the other startups you've been a part of, is there anything particularly different about leading a startup with AI technology versus another type of tech?

Steven Kim:
Yes. Previously, I was working in IoT, print on demand, and some more hardware-based startups. Here, it's very tech-centric and AI is moving at lightning speed, especially with GANs and everything else happening in this ecosystem. Things are moving really fast. So we have to be twice as fast to just keep up with everything. I always tell my team that speed is key for survival. Because the only thing that we have compared to big companies is that we can move faster, decide things faster, and work on things faster. If we lose that speed, we're left behind. Things can change overnight in this landscape. So we try to be on top of everything as much as we can. We are always paranoid, always trying to be aware of what's happening in the industry.

Sheikh Shuvo:
Fascinating. Well, taking a step back, tell us more about what your company does. I saw some amazing partnership announcements with Renesas launched recently. It sounds like there's a lot of interesting things going on.

Steven Kim:
Initially, we started as a more service-focused company, building very lightweight AI models that our customers could deploy at the edge, like in door locks, control systems, driver monitoring systems, intelligent transportation. What ended up happening is that we focused more on becoming a product-focused company by providing a product that helps other AI engineers build, optimize, and deploy AI models faster and better. This pipeline we built, we dubbed it as Nespresso, not to be confused with the famous coffee maker. It's designed to increase our productivity, so tasks that used to take us six to two weeks to build, optimize, and deploy AI models, we can now do in a few hours or a couple of days at most. Our target market is not the large multi-billion dollar market for large language models (LLMs), but rather the smaller niche markets that require computer vision and SLMs to improve customer retention experiences and others.

Sheikh Shuvo:
Got it. Is NetSpresso available as an open-source framework for use?

Steven Kim:
Currently, NetSpresso is available in three versions, one of which is a trial version. You can sign up for free, and we give everyone 500 free credits. With 500 credits, people can usually build at least a couple of models. We also offer free models on our website. So, you can go ahead, sign up, and try some of our AI models that we have already built.

Sheikh Shuvo:
Awesome. With so many different AI companies launching products right now, it's both an exciting and a scary time, as you're battling for the attention of many developers. Have you found a way to evangelize your technology within the developer community? Are there any particular strategies that have worked to earn the trust of technical audiences?

Steven Kim:
That's a great question. To be honest, I don't think we are really good at evangelizing this at this point. However, the way it works with the developer community is that we show results. We show products that work for them. Developers are no-nonsense kind of people. They want to touch and try themselves to see whether it actually works. We don't just claim things that don't work. So, bringing this kind of product and making sure that our benchmarks are consistent with what they can reproduce is really important, better than just trying to evangelize right now. We may not be very good at evangelizing, but I hope that through our work, the work itself speaks for itself and can be trusted by the developer community.

Sheikh Shuvo:
Looking more inwards then, at the team you've recruited, when you're hiring someone for your team, outside of being excellent technically, what are some other main skills that you look for?

Steven Kim:
Yeah, that's also a great question. When we look at a candidate, the main thing, of course, is whether they have the knowledge. But knowledge itself can only take a person so far. The person needs to know how to communicate with others and have experience in actually building something in the past. Otherwise, they can build something that works in the lab but wouldn't work outside of it. We are in the business of bringing everything outside the lab to go into production. We've had really talented PhDs and graduate students join our team, but if they don't master the skills to bring something from R&D space to the commercial space, they won't be able to contribute as much as others. So that's a key thing.

Sheikh Shuvo:
When recruiting talent, have you found any major differences between recruiting for your Korean team versus the one based in the U.S.? How do you attract candidates and what types of interview questions do you structure?

Steven Kim:
We don't actively recruit our talents in Silicon Valley yet, but we hope to in the future. In Korea, we try to build a people-first and innovation-focused culture at our company so that people who join us will be happy to work here, rather than working on something they don't believe in and are forced to work on. AI engineers are rare to find, so for us, building this kind of culture in our company to attract and retain these people is important. They need to be committed to our mission and work with us. That's the most important part.

Sheikh Shuvo:
Makes a lot of sense. Looking at your day-to-day operations, either for you or your team, are there any ways that you're using AI on a personal level?

—

Steven Kim:
We do have driver monitoring systems and access control systems that we have built and use in our daily lives. In the States, we don't have a lot of digital doors and smart doors. I live in an apartment, and that's something I would love to have. In Korea, you would have a facial recognition-based access control system for your entrance. Drive monitoring systems are also something we'd love to have here. We do have a commercial product in the market that you can buy to monitor your driving and ensure you don't get into any dangerous situations.

Sheikh Shuvo:
Nice, nice. Those sound like two great applications that I would buy as well. Now, the very last question I have for you is for someone just graduating school right now who's interested in getting involved in the world of AI. There are so many different ways to go about it. What are the questions that you'd encourage that person to ask to make the right decision as to which way or which company to go with?

Steven Kim:
The main question they should ask is whether the company they want to join is focusing on the same type of problem they're interested in. What's the problem they're trying to solve? Some people are fascinated with cutting-edge technologies but lose focus on how this will impact the market and individuals. They have to find the technology both useful and appealing, not just for the sake of technology, but for how it will impact actual lives of people. We've had engineers who want to apply AI to make places more secure, convenient, and safer for our community. That's why we focus on things like intelligent transportation systems to monitor vehicles and pedestrians at crosswalks. People are committed to saving lives because latency is key in these critical tasks. We sometimes have to stop them from overworking because they would stay up all night to fix issues and deliver these products to the market.

Sheikh Shuvo:
Well, it sounds like you're recruiting your team the right way. That's a good problem to have.

Steven Kim:
We have great people. The co-founders have a great network and were able to attract a couple of people initially. From word of mouth, they told other colleagues, and that's how we've been able to attract a lot of the talent from the universities.

Sheikh Shuvo:
Well, Steven, thanks again for being a part of this today. I'm excited to play around with NetSpresso eventually and see more of your tech in the market.

Steven Kim:
Thank you. Yeah, if you have any questions, just ping me.

Sheikh Shuvo:
Thank you so much. Thanks, Steven. Bye.

Steven Kim:
Thank you.