虚拟化研发工程师（GPU智算方向）-基础技术
Core
Develop and optimize GPU virtualization technologies (vGPU, passthrough, API forwarding) and compute pooling engines to enable efficient, isolated GPU resource scheduling for AI training and inference.
Role type
Senior IC GPU virtualization engineer (infrastructure)
Builds
GPU virtualization layers, compute pooling components, and HPC adaptation tools for AI workloads.
Domain
Cloud infrastructure / High-performance computing / AI infrastructure
Deliverable
production ML models | infrastructure
Required skills
C/C++ or Rust, Linux kernel internals, KVM/Virtio/VFIO, GPU virtualization (CUDA/GDS/GDR), PCIe/RDMA/CXL/NVLink protocols, heterogeneous GPU adaptation.
Preferred skills
GPU performance tuning, operator optimization, DPU/smart NIC development, contributions to Linux kernel/QEMU/KVM/Rust-vvm/Kata Containers, familiarity with TensorFlow/PyTorch/vLLM/SGlang.
Responsibilities
Implement and optimize GPU virtualization schemes (passthrough, vGPU, API forwarding); Design and build GPU compute pooling components for fine-grained resource scheduling; Optimize data transmission paths for AI frameworks to minimize performance overhead; Adapt and integrate mainstream and domestic GPUs (NVIDIA, AMD, Cambricon, Ascend) in virtualized environments; Troubleshoot and enhance stability of the intelligent computing platform in virtualized environments.
Seniority
Senior, hands-on IC
