Kotaro Tanahashi

Career Background

Recruit Co., Ltd. (formerly Recruit Communications)

Worked as a machine learning engineer in the internet advertising distribution division.
Built logic for real-time bid pricing and ad selection processing 50,000 requests per second.
Responsible for improving ad delivery algorithms, business applications of quantum annealing technology, and OSS (PyQUBO) development.
Found great interest in accelerating and improving inference accuracy in large-scale systems, and in refining logic through discussions with team members.

Why I Joined Turing

When ChatGPT emerged at the end of 2022 and I was convinced that LLMs would dramatically transform the AI world, I learned that my acquaintance Ochi was working at Turing. I gathered information through Twitter and became interested, but what attracted me most was the goal of achieving vision-based autonomous driving using foundation models such as generative AI. I decided to join driven by a strong desire to create a vision-based autonomous driving model from Japan that would astonish the world.

What I Do

Management of Autonomous Driving Group 1
Overseeing four teams — E2E Advanced Development, ScaleUp, RL, and MLOps — building and operating a development structure that leverages each team’s strengths.
E2E Autonomous Driving Development
Making development roadmap decisions and technical decisions toward achieving E2E autonomous driving, unprecedented in Japan.

What Makes My Work Rewarding

Taking on the unprecedented challenge of E2E autonomous driving development from scratch
The thrill of seeing real vehicles drive autonomously on public roads using models I developed
Cutting-edge deep learning development using large-scale GPU clusters
Experiencing daily model evolution through an overwhelmingly fast development cycle
Development spanning diverse areas including cameras, hardware, and training
Applying cutting-edge technologies like 3DGS and reinforcement learning directly to the real world
Pursuing the ambitious goal of creating world-class technology from Japan
Competing globally at the forefront of the autonomous driving industry’s intense technological currents

About the Work Environment at Turing

Well-equipped development facilities including large-scale GPU clusters and the Heiwajima Lab
A structure where specialized teams leverage their strengths and collaborate to solve challenges
A speed-oriented culture that embraces failure, running dozens of experiments and improvements per week
An environment where you can engage with a wide range of technical domains from hardware to software
High degree of freedom to pioneer unexplored technical domains

Who Thrives at Turing

People who own a car and love cars
People who can find solutions through trial and error for unprecedented challenges
People who can persevere through gritty debugging and repeated trial and error
People who can adapt to rapidly changing technology trends and continuously learn new technologies
People who can collaborate and develop across the boundaries of hardware and software
People who can work with a sense of speed in fast development and improvement cycles
People with the positivity to move forward in collaboration with the team even in difficult situations
People who can take interest in the full End-to-End process from data collection to model training
People with the ambition to make a significant social impact through their technology
People who enjoy the unfinished phase and can build environments and systems themselves

For Those Considering Turing

Turing is now at the phase where we have finally reached the starting line in E2E autonomous driving development. While we are achieving autonomous driving in Tokyo, the road here has been far from smooth — it has been a continuous series of unknown troubles and painstaking debugging.

Therefore, those joining us need the resolve not to lament an unpolished environment, but to carve their own path and enjoy the gritty trial and error. However, what awaits here is a stage for ‘mixed martial arts’ — equipped with large-scale GPU clusters, real vehicle environments, and professionals from every field. The excitement of seeing cars move with your own code and witnessing daily evolution is an experience you cannot find anywhere else.

For those with the strong conviction to create world-class autonomous driving from Japan and the ability to fight through intense global competition with overwhelming speed, there could be no more stimulating and exciting environment than this.

My Car

Audi S4

Why I chose this car: The fuel efficiency is not great and being a left-hand drive car makes it slightly inconvenient, but I was drawn to how it combines excellent ride comfort with sportiness, more than making up for those drawbacks.