Interview

I'm currently majoring in information engineering at a university in Nagoya. I'm mainly studying IT and programming, and I'll be a 4th-year student starting in April. I want to move into the AI field in the future, particularly to academically understand the mechanisms of LLMs and similar technologies.

I originally wanted to be involved in web development as a long-term intern. However, there aren't many web-focused companies in Nagoya... Since it's centered around automotive companies, there were hardly any environments where you could do pure web system development. So I thought about finding a company in the Kanto region where I could work remotely, and while browsing job sites, I felt that Samurai Technologies 'had exactly the conditions I was looking for.' What particularly appealed to me were these points: - Being able to be genuinely involved in web development - Being able to develop using cutting-edge technologies like generative AI - Being able to take on challenges even without development experience (beginners welcome) - Being accepted even as a student Job listings with all of these were practically impossible to find. So Samurai Technologies' posting was a perfect match in both conditions and content. I thought 'This is the one!' and applied immediately.

Yes, I was. It started because I had a friend working at a web development company who told me, 'Web development is fun and it can actually generate revenue.' Influenced by that, I started playing around with JavaScript and HTML. What's especially appealing about the web is that you can publish what you've made right away. Showing it to friends, making improvements. That casual nature was fun. But with personal development alone, it inevitably ends up as just 'self-satisfaction.' I started wanting to develop marketing perspectives too — 'How do you monetize this?' 'How do you deliver it to the world?' — and decided I wanted to gain experience at a company actually doing web development.

Currently, I'm mainly responsible for web system development. This includes development in cloud environments using GCP, building search systems, and implementing AI-powered features. I've already been involved in 3-4 product developments. Since I don't have much development experience myself, the basic flow is that Masuda-san comes up with ideas and I bring them to life. However, I don't just build everything as told — I also share my own opinions on specifications and features, and we move forward through discussion.

Since we work remotely and can't communicate directly, we keep Google Meet connected throughout working hours. Screen sharing is basically always on too. This creates a state where anyone can be spoken to at any time. With Masuda-san, we frequently have casual exchanges like 'What do you think about this?' 'How should we handle this specification?' My opinions are properly heard, so it's a very easy environment to work in.

The most difficult part is 'deciding specifications.' Generative AI provides significant support for the code itself. However, the design part — 'what rules should the system operate by' — must be decided by humans. For example, what rules to establish between users and providers, how to handle unexpected cases, and so on. These design decisions have major downstream impacts, so they can't be made easily. 'What happens in this case?' 'What about a different scenario?' Considerations keep piling up one after another. Once you start thinking, it truly never ends, and I stop multiple times wondering 'Is this really okay?' Honestly, it's tough, but proportionally, I feel a real sense of fulfillment knowing that I'm involved in building the foundation of the service.

Honestly, the evolution of generative AI is shocking. Development speed has increased incomparably from before. Implementations that used to take hours can now be shaped all at once. The barrier to 'just try building it' has dropped dramatically. As a result, I can now run the cycle of 'build → test → break → rebuild' at an astonishing speed. If we had to do this by hand alone, I think it would have honestly been spirit-breaking. But thanks to AI, I can take on new challenges and keep improving. I feel that the current development style truly couldn't exist without AI. On the other hand, what I'm most strongly conscious of is 'not becoming too dependent on AI.' Generative AI presents answers, but those answers aren't necessarily correct. They can be wrong, and the premises can be off. And when bugs or problems occur, it's humans who take responsibility, not AI. I believe that must never be forgotten. Even for things AI produces, I make sure I can explain in my own words 'why this design was chosen' and 'why this code works.' Even if I can't understand perfectly from the start, I make sure to understand eventually, even if it's after the fact. AI is ultimately a 'powerful auxiliary tool' — it's not something that lets you abandon thinking itself. If I stopped thinking, my own growth would stop too. That's why I'm always conscious of not becoming too comfortable with convenience.

Honestly, I think people who can only write code will have a hard time going forward. What matters is: - Understanding the purpose - Having a business perspective - Using AI effectively while not depending on it 'Why are we building this?' 'Who is it for?' I think the people who can keep asking those questions are the ones who will survive.

In the short term, I want to release the products I'm currently building and have many people use them. I want to release them while having not just a technical perspective but also a marketing perspective. In the long term, I want to be involved in AI and LLM research. And I want to continue system development that leverages that knowledge. While I'm still a student, I want to keep taking on challenges at Samurai Technologies. I want to seriously pursue interesting things and things I love.