[No. 198] Yoshihiko Suhara

I am honored to have this opportunity to write for the Keio University alumni newsletter.

After graduating from the Department of Industrial and Systems Engineering and completing my master's degree, I have been engaged in the research and development of artificial intelligence technology, moving between a Japanese company, an American university, and a Silicon Valley startup. I am currently working on natural language processing R&D at a San Francisco-based startup called Grammarly. I've realized that I have spent more than half of my professional life in the United States, and the experience of studying at Keio University has undoubtedly become the foundation of who I am today.

I was a member of the Faculty of Science and Technology Athletic Association American Football Team (at the time, President Kohei Itoh was our advisor), so I commuted to the Yagami Campus for a full six years, from my first year as an undergraduate to my second year as a master's student, which was earlier than other Faculty of Science and Technology students. I continued to commute to Yagami even after becoming a working professional in the doctoral program. During my time in the research lab, I spent most of my days there, outside of lectures, including weekends, making Yagami feel like a second home.

In the Department of Industrial and Systems Engineering, I belonged to the Sakurai Laboratory and was engaged in machine learning research. At the time, artificial intelligence was in its second "winter," and I was fortunate to be involved in the study of machine learning, especially neural networks, before it came back into the spotlight. During every progress report and discussion, Professor Akito Sakurai repeatedly asked me, "What insights did you gain from that?" from which I learned the importance of always seeking the essence of things. Also, through reading groups and study sessions in the lab where we presented the contents of papers and books to each other, I was able to acquire the attitude of mutual teaching and learning among faculty and students, as expressed in the Keio University motto of "learning while teaching, teaching while learning."

The Department of Industrial and Systems Engineering covers a wider range of academic fields compared to other departments. As an undergraduate, the breadth of subjects to study was vast, and I struggled with lectures, reports, and exams, but the experience of being exposed to such diverse disciplines during my student years was invaluable. I clearly remember Professor Takeo Gyotai, who retired at the same time as our graduation, saying at the Department of Industrial and Systems Engineering graduation ceremony, "What you have learned in the Department of Industrial and Systems Engineering is like the North Star. Whatever you do from now on, it will be your guide." It was only after I had been working for a while that I truly came to understand the meaning of his words.

I also had the opportunity to join "Angels," an organization that managed the Department of Industrial and Systems Engineering's servers, and I am deeply indebted to Professor Tadashi Iijima and Professor Yoshihisa Shinozawa. From these two professors and my seniors, I learned the fundamentals of UNIX systems and the responsible attitude required for computer management. I never imagined that I would continue to be involved in computer management for over 10 years after becoming a professional. When I had meetings about system design and operation with IT vendors in Silicon Valley, they were surprised to learn that I was not a professional infrastructure engineer. The foundation I built during my time in Angels has proven to be very useful.

After completing my master's degree, I joined the NTT Research Centers and Institutes and was assigned to a department conducting research and development in information retrieval. NTT had a training program designed to turn new hires into full-fledged corporate researchers over about three years, and I received generous guidance from my senior colleagues. At the same time, there was much to study on my own, and for the first three years in particular, I lived a life of working at the company and then staying up late to study after returning home. It's a fond memory to have organized reading groups at work and studied textbooks on information retrieval and machine learning with my colleagues.

In my work, I was able to engage not only in basic research related to information retrieval but also in the R&D of foundational technologies for commercial search engines and a map information retrieval service that became an Android app. During my time at NTT, they were very accommodating of my requests, and I had valuable experiences, such as conducting a research stay at the NTT Communication Science Laboratories in Kyoto.

In 2011, just as I was finishing my third year as a professional, I enrolled in the Doctoral Programs as a working student. However, my work became busier around the same time, and three years passed without me being able to complete my dissertation.

Amidst this, in 2014, I decided to change jobs and was assigned to a position at MIT. I managed to finish my dissertation in a rush and obtain my degree just before moving to the US. I have no words to express my gratitude to Professor Akito Sakurai, the sub-examiners, and the faculty of my major for their patient guidance and for accommodating my unreasonable schedule.

After successfully obtaining my degree and visa, I arrived in Boston in the middle of winter at the end of 2014. My advisor at the MIT Media Lab, Professor Alex (Sandy) Pentland, greeted me by saying, "Boston must be cold. It's the best place to do research." Boston winters are cold, but the buildings are well-heated and warm, making the indoors in a Boston winter the most comfortable climate in the world. (It was the author Mark Twain who said, "The coldest winter I ever spent was a summer in San Francisco.") The lab had students from diverse backgrounds, and I learned a great deal not only about research but also about different cultures.

Initially, my research at MIT did not go as planned. It was an interdisciplinary field called Computational Social Science, which uses computer science methods to tackle problems in the social sciences—an area I had never worked in before, so I didn't know the ropes. There were also language and cultural barriers, and for the first year, I produced no results at all. It was a really tough time for me mentally.

However, in my second year, I was able to publish a paper on a research project I was working on in tandem with a lab member. Once I got a feel for the research methodology, things finally started to click, and my papers were subsequently accepted one after another at international conferences and in journals. Still, the reality was that more of my research projects never resulted in a paper. But I learned the importance of not giving up and continuing to experiment even when results are not as expected, and especially in a new field, the value of working with a mentor-like member who shares a similar perspective.

Also during my time in Boston, at the invitation of a senior from the Faculty of Science and Technology Athletic Association American Football Team who was also in Boston, I completed a 5k-10k-half marathon medley for two consecutive years. The 5k race was held the day before the Boston Marathon and followed a course that crossed the marathon's finish line, so it felt as if I had completed the Boston Marathon itself, which became a very fond memory.

From 2017, I moved to Megagon Labs (then called Recruit Institute of Technology), Recruit's AI research lab established in Silicon Valley, to work on R&D in natural language processing and machine learning. We started with a small team of five, but it grew to an organization of over 20 people in five years. The team was incredibly diverse; at one point, the number of languages spoken exceeded the number of team members. Being a small organization, I was able to gain a wide range of experience, from discussing ideas across project boundaries to being involved in building the company's internal systems. I also increasingly took on leadership roles, mentoring interns and junior members in their research. It was around this time that I finally became comfortable using English for work and research. It took quite a while.

Last fall, I taught a course on Deep Learning for graduate students in the data science program at New College of Florida. Although I had experience giving lectures as a guest speaker, I was able to fulfill my dream of "standing on the other side of the lectern" as a faculty member for the first time—though it was on the other side of a Zoom screen, as the lectures were online.

Since this April, I have been working at Grammarly, a startup that develops a writing assistance tool, where I continue to conduct R&D in natural language processing. It reminds me of designing a language learning support tool in a small seminar taught by Professor Kyoko Ohara during my first year of university. It is deeply moving to think that I am now tackling the very problems I was thinking about 20 years ago. As Grammarly was founded by Ukrainians, many of my colleagues are Ukrainian, and they have taught me things like the Ukrainian way of toasting.

It feels like just yesterday that I was commuting to Yagami, but I realize that 14 years have passed since I finished my master's degree. And it seems I have spent more than half of that time in the United States. From my experience working in a Japanese company, an American university, and a Silicon Valley startup, I have come to think that although the language and culture are different, the essence of things may not be so different after all.

However, I believe it is important to understand cultural differences, embrace and adapt to change, and maintain a constant attitude of learning. In this regard, my studies in the Department of Industrial and Systems Engineering, with its wide range of academic fields and diverse student interests, have undoubtedly been a positive influence.

I have adopted the Keio University motto, "Practice makes the impossible possible," as my own. Through daily training, you make what was once impossible (or not understood) possible (or understood). I believe that sports, work, and research are all the same in this respect.