The Creator of [SpiderAF]: An Interview with CTO Eurico Doirado – Part 2 –

Today we sat down to talk with the leader of the engineering team, and muscle behind [SpiderAF] – CTO Eurico Doirado! In the previous entry, we talked about his background and his reasons for coming to Japan.

Now we’ll be going into what his impressions were like when he joined Phybbit, his thoughts on team-building, and more! Let’s start off by talking about how he learned about Phybbit!

My first impression was that this was going to be a “Crazy Adventure!”

Eurico when he joined Phybbit in 2013
Developing games was a fun experience, but I had very strong thoughts on creating my own business. After I left the previous company, I started to work as a freelancer with a designer to create games. It’s through that relationship that I got to talk to Ohtsuki-san (Phybbit CEO) about my ideas and she invited me to join Phybbit to work with her. That was April 2013, meaning that this was Phybbit’s 2nd year. My first impression was that this was both exciting and scary. It was scary in the fact that I couldn’t imagine where we would wind up. It was the same feeling as when I first came to Japan. “This was going to be one Crazy Adventure.” But it’s because I love that kind of feeling, I immediately agreed and ended up joining 2 weeks later.

Going from a data cleansing framework to an anti-fraud tool!

How did SpiderAF begin? Give us the inside scoop!

While we were working on B2C and C2C projects, we found a business opportunity in B2B. When we were in the process of helping several ad tech companies develop their services, we gained a reputation for great technical quality and speed. To address the common issue of reducing man-hours and labor, we developed an internal framework that facilitates data analysis, preprocessing, and machine learning that automated corporate workflows. We heard about issues related to ad fraud, but we never predicted that it would turn into a problem of this scale. Japanese ad networks have been investing in internal resources to fight fraud for a long time now. Although it was effective to some extent, we realized that advertisers, publishers, and everyone in the supply chain can only be effective in the long run if they worked together to fight fraud. SpiderAF was created at the end of 2016 to help ad network operators tackle ad fraud.

SpiderAF is tackling ad fraud of varying levels of difficulty and complexity, but how do you manage that with a small team of engineers?

Engineering Team
It’s not easy (laughs). Our clients include advertisers and publishers as well as ad tech companies and agencies. The technical labor to maintain a service like “SpiderAF” is substantial as the ad tech ecosystem is fragmented across different stacks of technologies. The challenges include:
  • Partner integration; management of many existing ad tech systems must be integrated to properly report fraud
  • Pre-processing, cleansing, and analysis of tens of TB of data per day
  • Not only quickly understanding all of the signals we discover, but also reporting details in a way that each stakeholder can understand
  • Research and development of fraud detection methods with the constant emergence of new fraud techniques
The new features we offer must be ready to work at a much larger scale than what most services can usually handle. Of course we don’t implement everything ourselves, for example we rely on Google’s BigQuery and Apache Flink for data processing and the Python ecosystem plays a substantial role in our analysis process. When it comes to our web dashboard and reporting stack, we turned to Elixir / Erlang as it provides great functionality out of the box.

How will SpiderAF evolve? How is the technology stack expected to change over the next few years?

I can’t comment much on current and future initiatives (laughs). One of the products we are shipping now is real-time fraud prevention. We provide white-box solutions to fraud and customers trust the effectiveness and transparency of our technology. With this product we are blocking fraudulent traffic before impacts advertising budgets and processes down the line.
Due to the many technical challenges of making real-time software, we redesigned some parts of the architecture. We now use Rust for performance-critical components, all of which are important when handling billions of requests. In addition, we also realize that brand safety in mobile applications is also a very difficult task. Unlike websites, mobile apps are built with stacks that third-parties cannot easily inspect. I’m looking forward to talking about this in the future.

On the lookout for engineers who are both generalists and specialists!

In order to expand services in the future, getting support from great engineers will be essential. Please tell us what kind of engineer you are looking for.

We are currently looking for software engineers who understands ad tech and cybersecurity. Required skills can be broadly divided into data analysis & security, systems, and UI / UX. 
  • Data analytics & security: research how ad fraud happens in the supply chain, finding new schemes, collecting and analyzing new data, creating detection & prevention techniques.
  • System Engineering: handle large amounts of data, large amounts of transactions/requests. Feels comfortable diving to a very low level into our stack and optimize it for speed, reliability or scalability.
  • UI / UX design: breaks down complex information and processes into easy to use and visualize reports. Can make it all look and feel obvious to our users.
Communication skills are essential to understand the needs of our clients and facilitate cooperation with other engineers. One trait that all our engineers have in common is that they can work pretty much on any part of the stack while being experts in some domains; they are both generalists and specialists. Finally we expect people on our team to be independent, to be able to manage the whole process of transforming a fuzzy idea into a fully featured functionality.

Managing the Japan and Portugal engineering teams as CTO

(Left: Eurico、3rd From left: Gonçalo. Climbing Mt. Fuji in 2011)

I think having a team of highly skilled engineers is a product’s strength. Starting in January 2019, I heard you set up a base in Portugal and that you are actively hiring local engineers to strengthen our products!

Yes. Gonçalo, one of the engineers at our Portuguese office, was a friend of mine from college and we worked together in Japan after graduation. After that he returned to Portugal, earned his Ph.D. and worked for another company for while. In 2018, we called out to him to start the Portuguese office and that’s how we ended up starting over there as well. In Europe, talented engineers gather up in London, Berlin, and Paris. But recently, Lisbon in Portugal has become one of the hottest cities to be in. However, local engineers are not properly valued and often do not get rewarded for their skills. Providing the same high-level environment and rewards as Japan is a huge advantage for everyone. JOIN US!
Would you like to join Phybbit and be part of a team with diverse values and backgrounds?