Welcome to our team profile blog post! In this article, we highlight Yuliang Li, a Senior Research Scientist at Megagon Labs. We’ll discuss the aspects of the lab Yuliang appreciates, how he arrived to Megagon Labs and words of wisdom to research scientists.
Tell us where your journey to NLP and database research began and how you managed to succeed?
I have studied computer science since middle school. I was in a Competitive Programming club in high school. I was actually not doing very well in the program when I started, but I continued to push through because I liked it. I didn’t get a significant award during high school, but by that time, my interest in computer science grew substantially. I went on to Hong Kong University of Science and Technology (HKUST) for my undergraduate studies. HKUST has a very strong focus on research and publication. They encourage undergraduates, even in their first year, to join research groups that conduct serious academic research under the supervision of PhD students. That’s how I went from being interested in CS to pursuing CS research. At the time, I worked with Professor Wilfred Ng and Dr. Qiong Fang, who guided me into database and data mining research. I collaborated with them on a project about mining order-preserving submatrices in various applications, including gene analysis and RFID tracing. Having this experience made publishing a big interest of mine.
During my undergrad, a major event was spending a semester studying at Rice University in Texas. At Rice, I took a very special course called Logic in Computer Science. The professor turned out to be Moshe Vardi, one of the top experts in that field. I learned quite a lot from his course about logic and how it applies to databases. For example, we learned about the connection between SQL and first-order logic. I did well in that course, and he actually wrote me a reference letter for my PhD application. I think it really helped me get into UC San Diego for my PhD. My undergraduate studies really built a good foundation for my PhD research on the intersection of logic and databases.
Tell me about your journey to Megagon?
My time at University of California, San Diego ended in 2018, when I finished my PhD in database and database theory. I was looking for a position where I could continue to do research. At the same time, I didn’t want to go into academia because it was very competitive and I wanted to do something in the industry which would have a real-life impact.
I found Megagon Labs because of Wang-Chiew Tan and Alon Y. Halevy. In 2013, I was a Google summer intern on Alon’s team, way back when he was leading structured data research. Wang-Chiew was the PC chair of the PODS 2016 conference, where I published a paper. Both helped me look into Megagon, and I found they were doing research in databases, data management, and data integration. I felt it was a very good match due to my research interests; it also seemed like a nice blend of academia and industry.
I submitted my CV, passed the interviews and boom, I got it. Megagon was my first choice, and I got the offer. It was a very happy moment for me.
What’s the most interesting thing that happened with your work at Megagon recently?
The most interesting part is collaborating with our interns on research projects. My peers and I have the flexibility to choose the topic of the projects and lead the research, so it makes for a great learning experience. For the group two years ago, I chose to have us work on meta-learning: a topic that was not very close to what I was working on at the time. I looked through the publications on meta-learning and AI, so I thought it would help our current projects, Ditto and Snippext. That summer was a very good learning experience for the interns and I. We read papers, dug into people’s code bases (even PyTorch) to understand what meta-learning is and how to implement and integrate it into one of our projects. – A big shout-out to our former intern Zhengjie Miao and my colleague Xiaolan Wang for the amazing work!
I liked this freedom of exploration—a quality you can still find at Megagon every day. We are allowed to dig into anything we find interesting, as long as it helps us advance towards our company goals. With meta-learning, we found ourselves making an impact by publishing new research and contributing to real-life applications.
Speaking of publications, something else I really enjoy and have fun with is naming our papers. We have been using Pokémon characters lately. For example, our latest paper on contrastive learning for data integration is named after Sudowoodo, a Pokémon that has a tree-like appearance and is hard to distinguish from a regular tree. This just shows how much we like our work.
What piece of advice would you offer to aspiring research scientists?
I would say this: Focus more on topics of research that are interesting to the community. There are many ways to publish papers, but I would say nowadays there are too many papers but not enough interesting papers. Instead of publishing a lot, focus on a few interesting topics or important challenges that people have been looking into for years. Make your research more impactful by trying to solve problems that have long been a challenge in the industry.
What is something you like about the Megagon culture?
I didn’t know that much about coffee before joining Megagon. I used to drink one-dollar coffee sold at the CS department at UC San Diego. PhD students need a lot of caffeine. I used to just bring my big mug and fill it up for $1. My knowledge of coffee was just that I needed it and I could get it for just a dollar.
When I interviewed with Megagon, we had lunch together—in the middle of the interview—and after that, we had a casual chat in the kitchen. They were making pour-over coffee, and they asked me if I wanted to help.
It was my first time making pour-over coffee, and I found it quite interesting. I learned about coffee. I wanted to join the conversation. After I joined Megagon, my colleagues were quite interested in sharing that coffee culture. They were eager to teach me and gradually, I learned what good coffee really is. I got so into it that I started to do my own research on coffee. Coffee roasting and grinding my own coffee became one of my hobbies. I compared the quality between my home-roasted coffee beans and store-bought coffee beans.
The part I enjoyed most then, and still do, is making coffee for the team. We keep improving our techniques. We have a spreadsheet and a controlled process.
From my coffee experience, I learned the team at Megagon is very friendly and very eager to share their knowledge. We always try to have fun in the office.
Follow us on LinkedIn and Twitter for to stay up to date with us.
Written by Natalie Nuño, Yuliang Li, and Megagon Labs