BLOG_LINE Fukuokaデータ分析image1

LINE Fukuoka has a team that specializes in data analysis where data and machine learning engineers as well as data scientists play an active role. In this interview, we asked them about their careers up until now, and what they are working on in Fukuoka.

Career History and Reasons for Joining LINE Fukuoka

―Would you tell us about your careers up to this point, and how you came to join LINE Fukuoka?Career History and Reasons for Joining LINE Fukuoka
BLOG_LINE Fukuokaデータ分析image2

Before I joined LINE Fukuoka, I was mainly creating Web services using Ruby and PHP. For a while, I was also making computer vision and machine learning-related products on the side for fun!
After joining LINE Fukuoka, I briefly worked on server-side tasks using Perl and Java, and I also developed some Android applications. Besides all of this, I put together several demonstrations showing what we could do using machine learning, made proposals to our planners, and actually integrated this technology into our system.
After gaining all of this experience, I transferred to the Data Analysis Team at its launch in October of 2016, and am currently in charge of machine learning and analysis-related tasks.

―How did you come to join LINE Fukuoka?
When I heard that LINE had a development center in Fukuoka, I simply thought it was interesting. I heard from an acquaintance who was already working as a LINE engineer that the company’s interview test for engineering positions was really difficult and it got me interested, so I decided to give it a shot and applied for a position. So that’s how I came to work here.
BLOG_LINE Fukuokaデータ分析image3

―Mr. Hongo, would you please tell us about your career before joining LINE Fukuoka?
I’m from Fukuoka, but went to Tokyo to look for work, so I have a variety of experience, but I worked the longest for a security company.
I did analysis of malware on virtual machines, but in terms of data analysis, I did a lot of high-volume malware implementation on virtual environments, taking the data and determining what sort of malware was prevalent and how we should go about detecting it. It was kind of like using data to categorize malware.
After gaining that experience in Tokyo, I developed a columnar database compatible with MongoDB that was capable of writing queries at a Canadian company that makes database products.
When I started thinking about returning to Japan, one of the first companies I was fortunate enough to encounter was LINE Fukuoka.

BLOG_LINE Fukuokaデータ分析image4

―Ms. Lu, you joined the company from abroad, didn’t you?
―Yan Lu
Yes, I joined the company last October. My previous job was at a Chinese telecommunications company where I did data analysis of broadcasting devices and routers.
I thought I might like to try SNS data analysis, but there was no such work at my previous job. At that time, I also didn’t really know the difference between Tokyo and Fukuoka. (Laughs) I learned about LINE Fukuoka from the recruiting representative, who told me a lot about both Fukuoka and the company’s work environment, and I decided to try working here. I had also wanted to try living in Japan since I was a child, so it worked out perfectly!

―Mr. Oshiro, what has your career been like up until this point?
I’m originally from Okinawa, but I found work in Tokyo as a Web engineer. After that, I got involved with ad analysis and DMP construction in the data analysis section, where I really started to delve into the world of data analysis.
Next I transferred to a venture company that a university friend of mine had set up, where I worked on data analysis for various clients for about a year and a half. That company now has around 100 employees, but at that time there were fewer that 10 of us, so I was doing everything from analysis to sales and even contract drafting! (Laughs)
I was getting a lot of valuable experience, but for personal reasons, I wanted to make Fukuoka the base of my life. I also felt like I wanted to try doing analysis and improvement proposal for company-owned services rather than just data analysis for clients, so I decided to join LINE Fukuoka with its vast amount of information as well as many related services.

BLOG_LINE Fukuokaデータ分析image11

The Launch of the Data Analysis Team and New Efforts Originating in Fukuoka

―What type of tasks is the Data Analysis Team doing?
After Fukuoka launched the Data Analysis Team, the first thing we did was ask for an explanation and status report of the systems that LINE Data Labs in LINE Corp. (Tokyo) had, and gradually adjusted to be able to do some of their tasks. We now communicate with the planners in Tokyo, and are working at the same level as Tokyo. In fact, the number of tasks that Fukuoka is overseeing has recently increased.

―What sort of tasks are you in charge of?
Everyone on LINE Fukuoka’s Data Analysis Team has a differing area of expertise, so the tasks we receive are totally different. My main task right now is drafting reports that can visualize in numbers the company’s monthly sales and how effective its ads are. LINE offers a variety of services and accumulates an enormous amount of data on a daily basis, so I need to shape my reports as easy-to-display data. I write the ETL process for it and make the finalized report easy to read and attractive when it is output. Besides all of this, I also have several machine learning-related tasks.

My job is to collect data such as user characteristics that planning managers reference when pursuing service improvement or new strategies.
Also, in a more infrastructural sense, there have been in-house requests to view collected data freely, so I look into how to configure servers when various people execute queries and resource management for the middleware side, or security for preventing people from seeing data that they shouldn’t.

I frequently work together with Ms. Lu on the Growth Hack project, which involves deep data analysis of LINE services in order to grow them.
Incidentally, the team I belong to is based in Fukuoka, but I often collaborate with Tokyo members in pursuing the actual work. In the Growth Hack project that I’m now mainly in charge of, I work with Tokyo’s LINE Data Labs data planners as a hub, having them connect me to various service-side staff and using our video conferencing system to share daily progress or conduct once weekly debriefings with the service-side, as well as taking once monthly business trips.
Even in a remote environment, we try to get creative by turning information into text on Wiki and chat rooms so that (to the extent possible) there is no gap in our understanding, or pair improvement proposals for service growth with analysis reports so that we do not simply end up with data.

When the Data Analysis Team was launched, we talked about wanting to be able to take on a variety of tasks unique to Fukuoka rather than just receiving jobs from Tokyo, so in that sense the Growth Hack project that Mr. Oshiro and Ms. Lu are working on is a good example something really unique.

It’s true that at this point in time, we’ve heard from LINE Data Labs that Fukuoka (and not the original planners for services we are in charge of) is the first example of a team delving so deeply into the analysis of LINE services.
Even before the Growth Hack undertaking began, we were receiving and responding to requests from each service for ad hoc or “shot” analysis (a type of short-term analysis), but the structure was basically us responding to the content of an order we received, and once we were finished, starting our response to another request for service analysis.
With the Growth Hack project, however, connections rapidly emerge between the original dots of our shot analysis content, since we continually track figures in order to achieve service KPIs. Because the project also progresses through the pairing of services and tags - from the upstream part of asking ourselves what we should analyze to the outcome of improvement proposals - from an analyzer’s standpoint, I think the places where we can feel a sense of unity with the services are interesting.

As Mr. Oshiro said, the project that two of us are now involved in is the first embodiment of our discussions about working with everyone from analyzers to service planners in LINE Data Labs on the Growth Hack project.

BLOG_LINE Fukuokaデータ分析image6

―What is the background behind doing the Growth Hack project in Fukuoka?
We received a request from the service side to work on the Growth Hack project together in order to improve LINE Family services. LINE Fukuoka’s Analysis Team was established at around the same time of that request, and it also coincided with an increase in our data scientists due to Mr. Oshiro and Ms. Lu joining the company, so Fukuoka’s Analysis Team responded to the request.

Yan Lu
I had experience at my previous company, but I want to be able to do more, so I’m studying as I work together with Mr. Oshiro on analysis. Mr. Oshiro and Mr. Morimoto from Tokyo’s Data Labs are planning to announce how the project is progressing at the “LINE Developer Meet up in Tokyo#18” to be held in July.

Data Analysis / Machine Learning Engineer Community in Fukuoka

―Are there many data analysis / machine learning engineers around Fukuoka?
Well, I regularly attend study sessions held outside of the company, and recently participated in an R language study session, but my feeling is that there are still fewer of us here than there are in Tokyo. People may be interested in and want to try it, but I get the impression that not many of them are actually doing it professionally.
Again, this is just my personal impression, but the atmosphere in Fukuoka now feels the same as Tokyo’s atmosphere did a few years ago when ”Big Data” and “data scientists” were all the buzz.

Hardly any companies in Fukuoka use Hadoop while operating it on-premise.

That’s true. There may be some that entrust data to things like Cloud-based data management services and operate it from there.

In Fukuoka, I think there are still many cases where companies store data in-house.

My previous company used a Cloud-based data management service for a while (I’m not sure if it still does).

Speaking to participants at data analysis infrastructure study sessions, I got the impression that almost everyone was using Cloud-based data management services or AWS. I think a lot of engineers would like to try on-premise [data management], but the reality is that there aren’t many environments in which it can be done professionally.

Even if they are interested, it seems like there are various obstacles to actually doing it professionally, such as a lack of original data or sufficient environment.

At a study session that I lead recently there were about 100 participants, but I got the impression that many of the participants from Tokyo were veterans. A lot of LINE Fukuoka’s engineers have also joined from overseas. Speaking of which, I think Ms. Lu is probably familiar with the state of things abroad; as an analyzer from overseas, how do you view Fukuoka?

Yan Lu
Right, a lot of people in other countries are doing data-related jobs, too, so if someone wanted to continue doing the same kind of job while living in Japan and learning more about it, I think they can feel at ease taking on that challenge at LINE Fukuoka: there are many highly skilled professionals from overseas working here.
BLOG_LINE Fukuokaデータ分析image7

What We Want to Do, and Motivation at LINE Fukuoka

―I understand that you all have differing areas of expertise, but what kind of things do you want to do in the future?
Since I have a system programming background, I think it would be great if I could commit to open source while getting my hands on a lot of things like distributed processing middleware (which I’m actually using now).

My strength is machine learning, and now that I’ve transferred to the Analysis Team, I think I would like to do a lot more related tasks.
There is a Machine Learning Team in Tokyo’s LINE Data Labs which mainly does machine learning-related tasks, but I hope that even in Fukuoka, I can take on some sort of challenge that will be valuable to the field of machine learning.

For me, its overseas projects. Analyzing globally released services on a world-wide scale means that it doesn’t matters whether that analysis happens at the Tokyo, Fukuoka, or some other overseas office.
Currently, about 10% of LINE Fukuoka’s total workforce is from overseas, and as is the case with Ms. Lu, many engineers have joined the company from abroad. I think we can utilize the cultures, practices, and mindsets of various countries that we learn from them in the analysis of globally released services.
From the opposite viewpoint, if we could also (for example) turn business improvement-oriented data analysis for the many contact centers in Kyushu into an example, then an “only in Fukuoka” data analysis genre may be established.

Yan Lu
Since I just joined the company and am still a newbie among the other members, I’m always thinking of simply continuing to improve my skills. I’m also really interested in the machine learning sector, so I hope to study it while working on my assigned tasks.
BLOG_LINE Fukuokaデータ分析image8

―I understand the Analysis Team is now looking for engineers; what kind of people do you hope will join?
For as small of a team as we are, our style is actually doing a lot of proposals and not just receiving and responding to assigned tasks. The work we should do isn’t mandated, so I think engineers who enjoy that kind of flexible environment while taking on new challenges would make a good fit.
When we say machine learning and data analysis, some people may get the impression that we do a lot of difficult work or that they would need highly specialized skills, but that’s not necessarily the case. We would like to work with people who don’t limit themselves by deciding that something is impossible right away just because it seems difficult, and who rather have confidence in what they can do and pursue it.

Speaking to engineers at data analysis infrastructure study sessions in Tokyo, there were quite a few who wanted to try operating Hadoop on-premise, so I think if people with that kind of challenger mindset were to join LINE Fukuoka, they would find a lot of what we do to be interesting.
Also, if an engineer suggests something that they would like to try, the open environment at LINE Fukuoka allows them to freely take on that challenge, so you can build a broad knowledge.

When people hear “LINE”, they may think of it on a large scale, but LINE Fukuoka’s Data Analysis Team is still small. Everyone has their own area of expertise and expresses their individual personalities, so I think we have achieved a nice balance.
However, since we are still a small team, I would love to have people who are hungry to try taking on challenges with a “start-up” mentality join us. I think at this stage, that kind of person would make the best fit.

BLOG_LINE Fukuokaデータ分析image9

The Data Analysis Team members will be appearing in the “LINE Developer Meet up in Tokyo #19” scheduled to be held in July.

<LINE Developer Meetup in Tokyo#19>
Date and Time: Saturday, July 22, 2017; 2 P.M. – 5:30 P.M.
Venue: 【Tokyo/Shinjuku】 LINE Corp. HQ
・The Data Analysis Teams in Tokyo and Fukuoka
・User Interest Extraction and Visualization Via Large-Scale Text Mining
・Lead up to Implementation of PySpark in Apache Zeppelin
・The Growth Hack Project Implemented through Tokyo - Fukuoka Collaboration
Entries: Now being accepted via connpass *Unfortunately, this event has reached its capacity.

We are planning various other such events in the future, so please join us next time!


LINE Fukuoka is always looking for engineers. Please check out our available positions and related links below.

▼Overview of Data Engineer / Machine Learning Engineer Recruiting

*Other Available Engineering Positions