Neural Network Performance Engineer
Core
Optimizing neural network inference performance for robot control policies to enable smooth, low-latency motion on cloud and onboard hardware.
Role type
Senior IC neural network performance engineer (robotics)
Builds
High-performance control policies for humanoid robots (HMND-01 Alpha)
Domain
Robotics + Deep Learning Inference
Deliverable
production ML models
Required skills
GPU architecture, custom kernel implementation, model quantization, deep learning profiling, PyTorch/JAX, Python
Preferred skills
Robotics or autonomous driving experience, VLA/VLM model familiarity, open source contributions, publications at ICLR/ICML/NeurIPS
Technologies
NVIDIA Thor, PyTorch, JAX, Python
Responsibilities
Analyze performance bottlenecks in model architectures, implement custom kernels to reduce memory throughput, quantize models with minimal quality loss, suggest architecture changes for better performance, deploy models efficiently on new hardware
Seniority
Senior, hands-on IC