Software Engineer, Drivers & Control Plane, Dojo

About the position

As part of a small, experienced team, you will have the opportunity to contribute to Tesla's cutting-edge Dojo Supercomputer, playing a key role in shaping the future of autonomous driving, Optimus, and real-world AI. You will play a crucial role in optimizing Tesla's neural network training by developing user-space and kernel drivers for our upcoming in-house custom-silicon supercomputer systems. You will be responsible for building and enhancing the drivers and control plane that power the Dojo distributed training system.

Responsibilities

• Develop and optimize device drivers to enable seamless interaction between software and hardware in the Dojo distributed system
• Enhance the reliability and performance of the entire control plane stack, from drivers to cluster monitoring and repair routines
• Work closely with researchers and Dojo software engineers to profile applications, identify bottlenecks, and improve the performance of application and hardware interaction
• Collaborate with the Dojo hardware team to understand the architecture of current custom silicon and propose driver optimizations and future hardware improvements

Requirements

• Degree in Engineering, Computer Science, or equivalent in experience and evidence of exceptional ability
• Extensive experience with C++ and C programming languages
• Strong background in user-space or kernel device driver development, especially in PCIe-based systems
• Excellent communication skills and the ability to collaborate effectively with cross-functional teams