Distributed Systems Software Engineer, Collectives
Core
Develop, test, and optimize collective communication and networking software for AMD GPUs to improve training and inference performance across multi-GPU and multi-node systems.
Role type
Software Engineer (Distributed Systems & GPU Networking)
Builds
Collective communication libraries and network software for AMD GPUs
Domain
Hardware/AI Infrastructure
Deliverable
production ML models
Required skills
C++, Linux, Computer Architecture, Networking, Profiling, Debugging, AI Model Architectures
Preferred skills
Collective Communication (RCCL, NCCL, MPI), GPU Programming (HIP, CUDA), Deep Learning Frameworks (PyTorch, JAX, vLLM, SGLang), Distributed Training/Inference
Technologies
C++, Python, Linux, ROCm, RDMA, NICs, PyTorch, JAX, vLLM, SGLang, HIP, CUDA, RCCL, NCCL, MPI
Responsibilities
Implement, test, and debug features in collective communication libraries; Profile LLM and generative AI workloads to identify bottlenecks; Run performance experiments across GPU generations and network configurations; Contribute to tools for performance measurement and regression tracking; Support model enablement across AI frameworks.
Seniority
Mid-level, hands-on IC
