Kohei Watanabe
Career Background
At a major telecommunications company, I spent nearly 12 years developing, operating, and planning domestic cloud services while also working on several large-scale GPU cluster projects.
After that, while working as a systems engineer at a foreign storage vendor, I was approached and decided to join.
Why I Joined Turing
While the large-scale GPU cluster projects I experienced at the telecommunications company were very interesting, I had lingering questions about whether the business could recoup such enormous investments. Meanwhile, as data volumes used in IT systems were growing massively, I was drawn to the importance and depth of storage technology and moved to a storage vendor. When Turing later approached me and I heard about the ambitious goal of fully autonomous driving, I became convinced that only a goal this large could make large-scale GPU cluster investments truly meaningful, and I joined.
What I Do
- Planning and strategizing Turing’s compute resource allocation
- Coordinating and negotiating with stakeholders for compute resource procurement
- Designing, building, and operating cloud/on-premises GPU computing infrastructure
- Troubleshooting issues that arise on compute resources
What Makes My Work Rewarding
Building infrastructure of this budget scale and workload with such a small team is an extraordinarily rare experience — one I believe will be the first and last of its kind in my life.
The closeness to users (ML engineers) — working right next to ML engineers and being able to feel their every move and the urgency of incidents firsthand is invaluable.
There is strong executive support — the CEO and CTO take the stance that ‘you can never have too many compute resources,’ and as long as there are no financial issues, they strongly support investing the majority of capital in compute resources.
The cycle of technical problem-solving — infrastructure continuously improves so ML engineers can run training stress-free, and you experience a virtuous growth cycle where bottlenecks keep shifting.
About the Work Environment at Turing
ML engineers sit right across from us, so they can casually come over whenever they have issues like computations not going well.
In addition to our own GPU cluster, we use diverse cloud provider environments including AWS, GCP, and GMO, selecting them based on workload characteristics.
The main protagonists at the company are the ML engineers, and infrastructure works tirelessly to figure out how to best serve them — essentially having end users right within the company.
The commitment to maintaining on-premises infrastructure in the challenging financial world of a startup is a testament to the company’s unwavering priority on compute resources.
Who Thrives at Turing
Turing’s goal of fully autonomous driving is in a turbulent phase globally. Those who constantly ask themselves ‘What is optimal now?’ and ‘What should we do years ahead?’ while observing ML engineers’ activities, and who have a strong desire to make it happen, are a great fit. For infrastructure engineers who want to gain this rare and exciting experience, this is the best environment.
For Those Considering Turing
The experience in this role is so rare it may be a once-in-a-lifetime opportunity, offering an incredibly rich and fascinating perspective.
We are scaling toward the single grand objective of fully autonomous driving, aiming for the largest infrastructure scale in Japan.
Within this momentum, the experience of steadily accumulating technical expertise is invaluable for future careers, and we want to welcome more team members who can share in this exciting journey.
My Car
Tesla Model Y 2025
I always thought car-sharing was the best option. But I felt it was not right to talk about the company’s mission and the products of the company we aspire to without actually driving one, so I purchased the newly released model. Owning a car has expanded my range of activities, and I feel the strength of the automotive industry every day.
Cars I’d Like to Drive
I saw the Honda PRELUDE at Japan Mobility Show 2025 and am very interested in it.
