Summary
Hang Zhang is a Principal Engineer and Head of Foundation Model at XPeng with a decade of experience building large-scale perception and foundation models for autonomous systems. His career spans research and production roles at Meta, Amazon, Cruise, and NVIDIA, where he led projects in hardware-aware NAS, mobile detection toolkits, and multi-task perception models. He has a strong open-source and tooling background—contributing toolkits like D2Go and GluonCV/AutoGluon—and a track record of shipping efficient architectures such as FBNetV5 and ScaleViT into production. Known for consolidating multi-view, multi-modality perception stacks, he blends deep academic training (PhD, Rutgers) with pragmatic engineering that powers real-world driving and AR/VR applications. An interesting throughline: he repeatedly turns research prototypes into production toolchains that optimize models across platforms and devices.
10 years of coding experience