NVIDIA 2027 New College Graduate: GPU Architecture Engineering - China
Core
Conduct quantitative studies of current and future GPU architectures; develop performance and functional models; analyze graphics and parallel-compute pipelines; identify bottlenecks and propose architectural improvements.
Role type
New College Graduate GPU Architecture Engineer
Builds
Next-generation GPU and system features, performance simulators, modeling frameworks, and validation platforms
Domain
Semiconductor hardware design, GPU architecture, computer architecture
Deliverable
production ML models
Required skills
C/C++, Python, Linux, CUDA, performance/functional modeling, trace-driven simulation, SystemC, ASIC design, RTL, random-test development, deep learning model development
Preferred skills
GPU programming and debugging, memory-system architecture exploration, performance profiling and optimization, simulation/emulation/silicon bring-up
Technologies
CUDA, SystemC, Linux
Responsibilities
Conduct quantitative studies of current and future GPU architectures; Build and enhance performance simulators and modeling frameworks; Develop pre-silicon programming and test environments; Create directed and constrained-random test plans; Explore novel GPU composition and memory-system capabilities; Profile system and application behavior to diagnose bottlenecks; Validate designs across multiple stages from architectural models to real silicon.
