2027AIDU-大模型Infra工程师(J99967)
Core
Build AI heterogeneous computing acceleration engines and optimize large model training/inference efficiency across CPU/GPU/FPGA/ASIC architectures.
Role type
Senior IC infrastructure engineer (AI compute)
Builds
AI heterogeneous computing acceleration engines, distributed training/inference systems
Domain
AI infrastructure, high-performance computing, heterogeneous computing
Deliverable
infrastructure
Required skills
C++, Python, computer architecture, heterogeneous computing platforms (GPU/ARM/MIPS/DSP), MPI, NCCL, RDMA, CUDA/OpenCL, LLVM
Preferred skills
Large-scale AI cluster software development, supercomputing system software development
Technologies
CUDA, OpenCL, LLVM, MPI, NCCL, RDMA
Responsibilities
Develop and optimize low-level operators, communication libraries, and compilation systems; Tune performance for CPU/GPU/FPGA/ASIC architectures; Design adaptive distributed systems for large model efficiency; Explore algorithm-compute co-optimization to reduce costs.