About the Team

We enable Physical AI to operate reliably in the real world. Our mission is to maximize inference efficiency in on-device environments while meeting accuracy and latency requirements under real-world constraints. By understanding the computational structure of VLAs, building dedicated runtime engines, and engineering the software stack from scratch, we shape the next-generation of serving systems.

Responsibilities

• Analyze the architecture and operations of generative models, especially VLAs

• Deploy and run models on edge devices, utilizing hardware accelerators such as NPUs

• Identify performance bottlenecks, then research and apply techniques to resolve them

Minimum Qualifications

• Proficiency in C/C++ or Python

• Fundamental CS knowledge and English proficiency to understand ML systems research papers

• A proactive approach to problem-solving

Preferred Qualifications

• Experience deploying and optimizing generative models on edge devices

• Experience profiling performance at kernel and hardware level

• Kernel programming experience on hardware accelerators such as GPUs or NPUs