2021.05.12 Hyundai Engineering 분량13min
Words that appear the most in the recent technology related news would have to be AI related technologies such as deep learning. AI, which is being quickly implemented in various industries, is leading us to a more convenient, smart world. We met with someone who was said to approach deep learning AI in a more fun way and furthermore reflect it onto actual work. It’s Senior Manager Kim Sang-ho at Hyundai Engineering.
Q. Please introduce yourself, and tell us about the work you do.
Hello, I’m Kim Sang-ho, Senior Manager leading the Smart Engineering Team at Hyundai Engineering. I help engineers in the field by solving with their problems with AI. For example, I implement or develop text and image recognition technology for employees who review tens or even hundreds of drawings or research ways that could reduce the error range of even a single digit for engineers who prioritize accuracy. To do this, I plan training programs and form AI community for engineers in the field while also developing systems for the engineers.
Q. We hear that you are a very well-known AI mania in the company. How did you first get interested in AI and how did you study it?
It’s an honor to be called an AI mania. I first came across AI when I began studying the programming language ‘R’ back in 2015, when my wife was processing eye-tracking device data while pregnant and I was trying to help her. At the time, I referred to a MOOC (Massive Online Open Course) platform ‘Coursera’ when studying ‘R.’ And it was when I coincidentally came across the Stanford professor Andrew Ng’s machine learning course that I really got into deep learning. Since then, I acquired a total of 4 certifications ？ one each year until last year, including the MOOC Udacity’s Deep Learning Nano degree, which was the first one. Nano degrees are non-degrees that take up 4 to 6 months which more than allowed me to learn what AI was, since I did not major in the field. And in 2019, I participated in the group’s big data boot camp, learned about data analysis for 6 months, and focused on analysis tasks, which made me to firmly establish the foundation.
Q. Some people say that doing what you love as a job is more difficult than you’d think. How is it for you?
That’s not my case. Implementing AI to my work made the 8 hours I spent at my office felt like 3, 4 hours. What Steve Jobs said at the Stanford graduation, ’Your time is limited, so don’t waste it living someone else’s life’ pretty much sums up the coordination of what you love and your work. Doing something you love as a job makes you lead a proactive life. What matters is not ‘where I work’ but ‘whether I’m doing what I love.’ Also, I truly believe that there needs to be a process where what you love and the direction of the company is coordinated. I ask my team members “What is it that you like to do? or How could you coordinate it to the company direction?” as often as I can to let others know how great it is to do what you love for work.
Q. What has changed in your everyday life since becoming an AI mania?
Studying AI after coming home from work led to several burnouts. Naturally, I distanced myself from AI intentionally outside work. And focused more on my family. For example, cooking or working out with my family. But some time ago when I was spending time with my family, I saw my kids with playing cards. There were gods from the Greek and Rome myths on the cards ？ it was a simple game where you compare each character’s ability and characteristics and see who’s stronger. I coded the game with the kids and made it into a simple computer game. Each character’s ability was transferred into data, and you could compare them to see who wins. There was not much to it, but as it was a game that they liked, the kids had fun coding. I guess I really am an AI mania since I apply coding and data processing in my everyday life.
Q. It couldn’t have been easy to form a community with just a personal interest in AI; how did you come to found a community?
Lee Se-dol losing to AlphaGo in 2016 was what triggered the formation of a community. I felt threatened at the thought of AI related technology developing more and getting rid of all the engineers in the world, but I realized that engineers should not be dragged by technology but actively pave the way in the AI sector. So, I knocked on the doors of our company’s ‘Seed’ system. Seed is a program that supports every Hyundai Engineering employee with an idea; if the related department evaluates the idea and it is good, the idea proposing employee will be selected to take part in a project. I proposed the task to be carried out and the community idea which was evaluated favorably. And with the writer lecture in 2017, our community was officially established.
Q. What performances has the community achieved so far?
Externally, we have held various events like writer lectures and open houses through the community activity and changed the Hyundai Engineering employees’ recognition of AI; that it’s not something to be wary about but something that will make our lives better. When the AI community was first created, some made jokes about it being the avian influenza, and had no big expectations. Currently, our engineers bring up many AI related ideas themselves and are trying very hard to implement them onto our actual work. One of my personal feats is working with good people who are interested in AI. Seniors including Kim Gyeong-in, Lee Yoon-sik, Jo Jeong-won, Kim Joon-woo, and Jang Gyeong-sun are very active in the community as Managers. Also, everyone’s work is directly or indirectly related to AI in respective areas. It’s a privilege and my asset to be working with people who feel the same joy for the same topic.
Q. Would the AI-based plant automatic design system that you applied for patent last year have something to do with the AI community?
The AI-based plant automatic design system began as a technology development task in 2019 for Senior Manager Jo Jeong-won, one of our community members. Through the automatic design system, if you input the basic design conditions, you will have the structure modeling of a whole building complex within 10 minutes while cutting down the design cost by more than 20%. The level of innovation for such automatic design, data expansion, and optimized design was recognized, and we were able to have it patented not just in Korea but also in America, internationally. Recently, we have signed a technology agreement with Bentley Systems and are expanding Hyundai Engineering’s original technology to various areas including architecture, landscaping, plant, etc.
Q. We heard that you went beyond your community and now had your own team. Tell us about the Smart Engineering Team.
The Smart Engineering Team assists engineers in the field to easily implement AI and machine learning in their work. Our team takes care of reviewing tens and hundreds of drawings through image recognition technology, summarizes and organizes major keywords and requests of thick bidding documents through natural language processing technology, optimizes numerous scenarios that could take place in many ways quickly, and creates or predicts value though data analysis. Ultimately, we assist in creating an environment so that all engineers can easily approach the above-mentioned works and implement them onto their work.
Q. How could the Smart Engineering Team positively impact the Hyundai Engineering business?
Once the use of AI in engineering field, pursued by the Smart Engineering Team, is activated, we will first be able to handle simple repetitive work quickly, enhancing work efficiency. Productivity will increase as we will be able to automatically produce the results as well as accuracy, through the automatic review. Therefore, engineers will buy more time for digitization and focus on creating and analyzing data for more important decision making. Such positive cycle will allow Hyundai Engineering to be more competitive in the market.
Q. What is the team’s goal for this year?
This year, the Smart Engineering Team supports smart technologies out of all engineering center business plans. We are preparing so that the utilization of automatic design, image recognition and natural language processing technology could expand into purchase, construction, and test drive, based on AI technology design. We are also creating a digital platform that provides the environment for internal and external engineers, affiliates, and clients. Our team’s goal is to create a base which allows commercializing and providing various engineering services that make use of cloud native (DevOps, CI/CD, Microservices, and Containers).
Q. Many people who have not majored in the area are taking interest in deep learning technology as it is making headlines. What are your thoughts on this phenomenon?
I don’t think everyone has to learn the theory and algorism of deep learning and code models. Deep learning technology is surely leading innovative development in areas like image recognition and natural language processing which are applied onto our everyday lives, but people have to know that this technology could be unnecessary for their work. Once you understand machine learning in a larger sense (AI is a part of machine learning), you will be able to approach the concept of AI innovation that’s currently taking place a little easier. Also, as using RPA (Robotic Process Automat) or high functions of Excel (Power Pivot, Power Query, etc) allows you to get into the machine learning without coding, you will find a path if you are interest and seek it out.
Photography by. Jeon Seok-byeong