Alumni Voices: Kenny Song, Solving a Global Problem by Founding a Tech Startup in Japan

Kenny conducted research on adversarial machine learning and related topics at the University of Tokyo as a MEXT scholar after moving to Tokyo from San Francisco two years ago. He co-founded Citadel AI, a Tokyo-based tech startup that helps clients using artificial intelligence (AI) monitor and test their AI applications.

​​*This article has been edited from the original interview for brevity and clarity.

Please introduce yourself. What did you do before you came to Japan?

I was previously working at a large tech company in San Francisco but decided to move to Tokyo for my partner. My partner and I were in a long-distance relationship for several years even before I moved to Tokyo, so she was my main reason for moving here!

Also, it was a nice change of pace to study at the University of Tokyo, where I have the freedom to purely focus on projects that were interesting to me personally. During my time at the university, I began exploring startup ideas and found a co-founder. We raised a seed round of JPY 100 million in 2021, after which I left university to focus on the company full time. 

What is your company? What does it do?

The main product of Citadel AI is a monitoring software called Citadel Radar, which automatically detects and mitigates problems with machine learning (ML) models. These models are quite fragile in the real world because of operating issues such as data drift, model drift, and data downtime. 

The end-users of Citadel Radar are data engineers and ML engineers who want to improve their AI applications. Because the technology behind Citadel Radar is fairly vertical-agnostic, Citadel AI works with clients across many industries. 

Citadel AI currently has a team of four. As mentioned previously, we raised a JPY 100 million seed round in 2021, so we’re growing our engineering team. If you would like to join Citadel AI and help build the future of AI systems, please reach out – more info about job opportunities here!

What made you start your own company in Japan?

I’ve been interested in startups since high school, probably after reading the writings of Paul Graham, a renowned computer scientist, tech entrepreneur, and venture capitalist. Before coming to Japan, I’ve also founded and worked with startups in different countries, including the US, China, and the UAE, so it was a fairly natural decision to start Citadel AI here. 

The problem we’re solving, ML reliability, is deeply technically interesting to me, and I also believe there is a huge market opportunity to build new software foundations for ML applications. As society grows more reliant on automated systems, it’s important that companies can reliably build, verify, and deploy those systems. 

How has it been running the startup? What are some of your main difficulties?

Running a startup is fun and demanding. 

Startups are an incredible learning opportunity since you have to learn how to execute well across the board, whether it be engineering, product development, sales, hiring, or fundraising. There’s no other job in the world that you’ll work on all of these things at once. 

We’re lucky to have found eager early adopters for Citadel Radar as well as supportive investors and partners, but hiring our first engineers to develop the product further is currently a big-time investment. It takes time to find the right people who are world-class talented, ready to join a startup, and speak English, especially with Japan’s COVID entry restrictions, but we’re making progress. And yes, we’re hiring!

What are the business opportunities you see in Japan going forward?

Locally, we want to help even more companies in Japan successfully deploy AI applications, and contribute to the startup and ML communities here. As part of our efforts to help the wider community, we recently started an MLOps engineering blog.

But we don’t necessarily view ourselves as a “Japanese” startup, even though our headquarters are in Tokyo. We’re solving a global problem and we have a global team, so we are pursuing customers both inside and outside Japan. 

Any advice for people who are thinking of starting their own companies in Japan?

Do it! Japan needs more startups in general, and more foreign founders in particular. A massively underrated driver of the US startup ecosystem is the role of immigrants – more than 50% of all unicorns in the US were created by immigrant founders. 

Most Japanese startups only target the domestic market, but if you’re an international student in Japan, one of your competitive advantages is your cultural background and global mindset. Use that to create the next globally successful unicorn! 

Edited by: Xiaochen Su, Ph.D., the Managing Director of the Study Abroad Research Institute, a non-profit organization seeking to promote study abroad in Japan.

Leave a Reply