Mercari has grown to become Japan’s largest marketplace app, where anyone can buy and sell. However, our journey does not end there. To fulfill our goal of creating a “global marketplace,” we have welcomed talents from all over the world to Mercari’s Tokyo office—around 40 countries and counting.
In the #WeMakeMercari series, we would like to introduce you to some of the members in the company.
At Mercari, Machine Learning (ML) plays an enormous part in improving and innovating our user experience within our C2C marketplace application. In this article, AI team engineering manager Yusuke Shibui (@shibui.yusuke) interviews fellow AI team member Antony Lam (@antonylam) for a glimpse into what it’s like to work at Mercari as an AI engineer.

Why did @antonylam, a former assistant professor in computer vision (image processing) at Saitama University, decide to work as a Mercari engineer? Read on to find out more!

Becoming a Mercari Software Engineer from an Assistant Professor at Saitama University

@shibui.yusuke:Hi Antony. I heard you were a university professor before joining Mercari. Could you tell us about your career so far?

@antonylam: I was a doctoral student in computer science at the University of California at Riverside in the US and during my studies, I participated in the JSPS Summer program at the University of Tokyo. This program allows overseas students to work at any research institute in Japan.
After the summer program, I returned to the US and completed my doctorate, but recalling my positive experience at the University of Tokyo, I decided to work in Japan again. I ended up working at the National Institute of Informatics (NII) as a postdoc, and later became an Assistant Professor at Saitama University, specializing in computer vision.

@shibui.yusuke: What brought you to Mercari? Did you know of Mercari before you joined?

@antonylam: To be honest, I didn’t (laughs). I got a message from a recruiter and I was thinking of changing jobs around that time, so I decided to listen to what they had to say. After a few interviews, I deepened my understanding of Mercari’s business. I read a lot of Mercan articles as well; almost all of the English articles, and even the Japanese ones with Google Translate.

Mercan editor: I’m so happy to hear that… (tears of gratitude)

@shibui.yusuke: Why did you decide to move to Mercari?

@antonylam: I thought the work-life balance would be good. To be honest, people can be quite overworked in the world of academia. I also felt like I could relate to Mercari’s “Go Bold” value, which encourages employees to take on new challenges.
I was also drawn to the fact that the company invests a lot in communication between employees. For instance, the shuffle lunch system, where employees are randomly broken into groups of four to go out to lunch together, regardless of team or position.
I especially like that the company officially supports participating in conferences.
But above all, I was interested in contributing to the development of a product that utilizes a vast amount of real-life data at a private enterprise.

Mercari AI engineer, @antonylam

@shibui.yusuke: And so you decided to join Mercari. Are you enjoying it so far?

@antonylam: Yes! There are a lot of multinational and diverse members, and we are proactively starting to work on D&I (diversity and inclusion) as a company. I don’t know about other teams, but I don’t feel any issues with language barriers, as the AI team basically does all their work in English. I also think the GOT (Global Operations Team)’s support is great. However, I will say that there are some old documents only available in Japanese.

@shibui.yusuke: I recall that soon after you joined, you were assigned to a project to develop a shipping method estimation system using machine learning. This feature allows us to suggest the appropriate shipping method based on the item information entered by the seller. I think this project was technically outside of your field, but thanks to your work, we were able to release it this September.

@antonylam: Yes, my specialty is in computer vision (image processing), and we used natural language processing for this project. So it was not my specialty, but it is still related to machine learning. Out of the many machine learning algorithms, I decided to use the very classic “Naive Bayes”, which is often used for tasks like spam classification. To be honest, I hadn’t used Naive Bayes since I was a student but I still remembered how to use it (laughs). In the end, I was able to achieve high accuracy despite the simplicity of the approach.

@shibui.yusuke: I think it’s great that you were able to complete the project. I also know that you are helping other team members by reviewing their academic paper submissions. What are the challenges of working as a machine learning engineer at Mercari, and how is it different from the world of academia?

@antonylam: One big difference is that we have enormous datasets thanks to our customers. But the data can be quite messy. This is because the data labeled by some of our customers is not necessarily correct.

Mercari AI team engineering manager @shibui.yusuke （left） and @antonylam

From the left Mercari AI team members ＠shibui.yusuke, @antonylam, @y-oshima, @cournape

We’re hiring! Related job postings

Profiles

Antony Lam (@antonylam)

Computer Vision Engineer in the Mercari AI team. He completed his PhD in Computer Science at the University of California, Riverside in 2010, became a postdoctoral researcher at the National Institute of Informatics (2010), and then an Assistant Professor at Saitama University (2014). He has 15+ years of research experience in Computer Vision and ML, has published at well-known venues (CVPR, ICCV, ECCV, PAMI), and was recently awarded his first patent in remote heart rate sensing via RGB camera. His background includes, computational imaging, hyperspectral imaging, fMRI analysis for neuroscience, image-based recognition, and ML in general. He joined Mercari in 2019 and is enjoying the challenges of contributing to industry while maintaining an active role in academia.

Yusuke Shibui (@shibui.yusuke)

Engineering manager for the Mercari AI team. Worked in several companies as a cloud, infrastructure, backend, ML engineer, and joined Mercari in July, 2018.