The company develops humanoid robots for real industrial environments - manufacturing, logistics, warehouses, and large-scale physical operations.
What makes the company especially interesting is that its work has already moved well beyond "research-only robotics." The company's humanoid platform is already running in real industrial pilots, it has roughly 30,000 global preorders, and its partners include major industrial and technology corporations. The company also collaborates with leading players in advanced computing and consumer electronics.
This role sits within the VLA (Vision-Language-Action) team - the group responsible for robot manipulation, autonomy, reasoning, and sim-to-real learning systems. In practice, the team builds the core intelligence behind how humanoid robots understand and interact with the physical world.
About the Role:
The role focuses on optimizing neural-network inference for low latency and high-frequency robot control - both onboard and in cloud environments - enabling robots to react safely and smoothly in real time.
What You Will Do:
Analyze performance bottlenecks of a particular model architecture and come up with potential improvements.
Make the model run on a new hardware (e.g. NVIDIA Thor) efficiently.
Implement custom kernels to reduce memory throughput requirements where it matters.
Quantize a model with minimal loss of quality.
Suggest and implement changes of model architecture that will enable better performance characteristics without sacrificing model capabilities.
What We're Looking For:
Experience building deep learning systems in research or industry.
Experience working on performance of neural network inference (analyzing bottlenecks, writing custom kernels, quantizing models, fighting deep learning compilers).
Excellent understanding of GPU architecture and why some models run faster than others.
Profound Python + PyTorch/JAX; you can profile, debug numerics, and write maintainable research code.
You document experiments clearly and communicate trade‑offs crisply.
Nice to have:
Robotics or autonomous driving experience.
Open source code showcasing your ability to improve inference performance.
Publications at ICLR/ICML/NeurIPS or equivalent open‑source contributions.
Familiarity with vision-language (VLM) or vision-language-action (VLA) models.