Trust and Safety (TnS) refers to monitoring, detecting, and handling prohibited behavior, including fraud, to ensure that Mercari users have a safe and secure experience. TnS in Mercari is not only implemented by the Customer Service (CS) Team which detects and handles such matters, but also by the TnS Team which is developing a platform that makes use of machine learning.

The TnS Team’s platform development started in 2018. What about the members of the team who have been involved in that work since its early days? What goes through their heads when they are taking on their tasks every day?In today’s article we present to you an interview with Suganprabu Nagarajan (@suganprabu96) and Nikinsha Kesharwani (@nikinsha), two engineers who have a long history with the team and who have supported the company’s expansion through their work. The interviewer is Engineering Manager Kenta Shimada (@kentan).

Didn’t know about Mercari before joining

@kentan: Let me first ask about the time before you joined Mercari. Suganprabu, you were a university student, yes?

@suganprabu96: Yes, I was studying in India. I was hired by Mercari during my last year at university, and joined as a new grad afterwards. Mercari was recruiting in my school, which gave me the opportunity to find out about the company in the first place. It was thanks to the information session and the hiring team members who explained about the company that I was able to get to know the basics about Mercari.

@kentan: You must have had other options than Mercari, so why here?

@suganprabu96: This was back in 2017, when Mercari was a small startup that wasn’t a listed company yet. I wanted to work in a company at that exact scope, so it was perfect timing for me. And I was also looking to work outside of India.

Most importantly, startups of that scale don’t usually go to universities overseas to recruit people there. If it was a company with an abundance of money and resources, that’s another story. But long story short, Mercari left a strong impression on me. There were many, many students who joined the information session to listen, which was also remarkable.

Suganprabu Nagarajan（@suganprabu96）

@kentan: How about you, Nikinsha?

@nikinsha: I was studying for a BA in computer science in India. After graduation, I worked as a backend engineer in an Indian tech company. Their tech stack was different than what we use in Mercari, actually.

@kentan: And you transferred to Mercari after your experience there. What led you to do so?

@nikinsha: I wanted to work in Japan…but I had no Japanese skills. So I was looking for a Japanese company where I would be able to work in English, and that’s how I stumbled upon Mercari. I discovered the company just as it was listed on the Tokyo Stock Exchange in 2018 and had a lot of traction as a unicorn. I thought it would prove an interesting experience, so I applied for a position.

Initially when I joined I did feel the language barrier, but soon after it was gone thanks to the language support provided in the company for English speakers. And on the work side of things, everyone has a strong sense of ownership when taking on their tasks, which makes it a fun experience as well.

Nikinsha Kesharwani（@nikinsha）

The work of the TnS Team, crossing over multiple domains

@kentan: We in the TnS team develop things to provide a safe and secure experience to our users. Suganprabu, you’ve been in this team for a long time, right?

@suganprabu96: Right. My main duties in TnS now are the development and operation of the moderation system. This system allows us to detect items that infringe the Terms of Service. Our team members go beyond their expertise and work across the domains of frontend, backend, machine learning, and more.

@kentan: You yourself joined as an ML engineer, but are now involved in frontend and backend as well. How’s it going with those?

Kenta Shimada (@kentan)

@suganprabu96: To go into a little bit more detail, I was an ML engineer until 2019, but have been involved in backend as well since then. As an ML engineer I had been doing development somewhat close to backend anyway, which meant I didn’t feel particularly out of my element. And starting 2021, I am also working on frontend. I am confident in my ML and backend skills, but frontend is new to me so I am studying to catch up!

@kentan: Isn’t it hard handling three areas at once?

@suganprabu96: They each require a different set of skills…so I guess it is hard in that sense (laughs). With ML, you have to be up-to-date with the latest research papers and literature, and with backend you have to be knowledgeable about APIs and architecture. Meanwhile, some areas of frontend are inseparable from UI/UX. I personally don’t find UI to be my strong suit, so thinking about it can sometimes be taxing.

@kentan: Nikinsha, what are your duties?

@nikinsha: As an engineer, I am working on the frontend/backend development of every project owned by the TnS Team. I am also the scrum master of the team, so I am responsible for improving processes and productivity. Like Suganprabu, I handle multiple areas at once for my development work, but I really enjoy being able to involve myself in all sorts of tasks.

@suganprabu96: I agree with that, actually! I am also working as the tech lead of the TnS Team. One of my responsibilities is to go over the rough plans of the features we will build during the next year and detail the technical limitations and issues we might possibly face. Sometimes there are multiple teams developing at the same time, so it’s also important to make sure everything works well together. I also review my team members’ tasks.

@kentan: Nikinsha, you joined as a backend engineer. What are your feelings towards working on other areas?

@nikinsha: I am right in the middle of my studies, actually. One of the unique things about TnS is that you can learn everything about the entire development cycle. Frontend, backend… As long as we’re handling it together with my team members it’s all fun work. Especially because frontend is a new area for me, I am very very interested in it and have a lot to learn!

Fraud detection has been a hot topic for the past few years, but how is it changing?

@kentan: I’d like to ask a few things about something very important for us at the TnS Team, which is safety and security. Do you sense any changes in that aspect between the time you first joined and now?

I personally feel that over the past few years, the work we do in the TnS domain, including fraud detection, has been garnering more and more attention in society as well. In addition to handling essential problems like infringements on the marketplace, now our work has more emphasis on Mercari doing what is right as “a public entity of society.” By having dialogue with not just our users but the companies who manufacture products and/or sell them to the public, we are able to maintain our position as a service that continues to grow and reach more people. We used to just delete every infringing item, but now we are taking measures like displaying warnings to users searching for items.

@suganprabu96: Yes, I think there has been a larger shift too. In 2018, it was my understanding that our TnS work was a “must” for providing a safe and secure service to people. Now, it is not just the TnS team that is taking on such work, but almost all of the company, as a part of improving the user experience. One of the changes that came along with this shift is that we are taking more efficient measures now, such as detecting infringing items more directly and speedily through machine learning.

The TnS Team until now has been individuals taking on different tiny tasks one by one. But now we are a feature-led team where members collaborate with each other and share a common objective.

@nikinsha: The Moderation Tool is also more effective now. It was a rule-based detection system before and wasn’t very efficient. It has been improved greatly since, which is a great change.

@kentan: The interesting thing about the TnS Team is that we are involved in both the CS side as well as the user-facing service. TnS has ownership of the internally-used tool, so we can make decisions on its architecture. We also have to develop its user-facing features in a way that can handle the immense traffic on the marketplace. Being able to see Mercari from these two perspectives is part of the fun of working in TnS, and our advantage.

@suganprabu96: Mercari uses a shared tech stack, so we probably aren’t very different from the other teams when training the ML product or service. But, because we have access to very special data collected through moderating the items on the marketplace, the models and techniques that can be built with that are very different than that of other teams.

@nikinsha: Most importantly, TnS is a cross-functional team. It’s very fun to develop many different things. Sometimes it gets hard, but I think I am thankful for the opportunity to take on challenges. And this is not about engineering, but there are many diverse teams in the company, which gives us the chance to learn about each other’s cultures and deepen our communication.

@suganprabu96: I also think the advantage of being in the TnS Team is the various challenges that come with it. Doing development work across multiple domains, you come across a lot of areas that you need to study. When there is a failure, you investigate the cause, take countermeasures, and move on. This is how you learn new things. Our team has members coming from a range of backgrounds, so there is always someone to consult on something. Internal tools, user-facing features, a high-traffic service… There are many different things we work on, which can be tough sometimes, but it means this is an environment where you can always challenge yourself.

Who’s a good fit for Mercari’s TnS Team?

@suganprabu96: I mentioned that I joined Mercari when it was still a startup, but I can’t really say that it has completely transformed into a “big corporation” either. Which means Mercari has that much room to grow! While we are much larger than a startup, we are a friendly organization open to flexible working styles with speedy progress and decision-making. All these elements coming together in one company isn’t common at all.

@nikinsha: Yeah, I agree. I think we are in a nurturing environment because it’s a company that keeps growing, in turn giving us many opportunities to grow alongside it. We got lucky finding this company.

@suganprabu96: I agree! On top of that I am also very grateful for the language support the company provides.

@kentan: It makes me so happy to hear that! Let me use that to lead into my next question: what type of person do you think would be a good fit to work in Mercari’s TnS Team?

@nikinsha: I’d like to yell from mountaintops that they need to be motivated to work with a diverse tech stack!Mercari has microservices for each of its major features, and operating those services is a big task. People who are willing to take on such a challenge would be a good fit for TnS.

@suganprabu96: I also think that people with multiple interests would enjoy TnS work. We don’t require deep expertise or experience in a certain domain. It’s more about being open to and hungry for learning new things. Also, because most of the moderation system is developed by the TnS Team, someone with a sense of ownership for the product would be great.

@kentan: I agree with you both. We’d love to add more people to our ranks who are curious and would enjoy learning and playing with multiple domains including frontend, backend, machine learning, and more.