Sr. Staff Distributed Systems Software Engineer, Collectives
Core
Design, implement, and optimize collective communication libraries and networking software for AMD GPUs to enhance training and inference performance across multi-GPU and multi-node systems.
Role type
Sr. Staff IC distributed systems software engineer (GPU networking)
Builds
Collective communication libraries, performance tools, and optimized distributed AI workloads for AMD GPU clusters
Domain
High-performance computing, GPU networking, distributed AI training/inference
Deliverable
production ML models
Required skills
C++, Python, Linux, collective communication libraries (RCCL, NCCL, MPI), network/NIC/GPU hardware architecture, distributed AI workload profiling, performance tuning, test design, mentoring
Preferred skills
PyTorch, JAX, vLLM, SGLang, RDMA, scale-up/scale-out network design, large-scale cluster optimization
Technologies
ROCm, Linux, C++, Python, RCCL, NCCL, MPI, PyTorch, JAX, vLLM, SGLang
Responsibilities
Design and optimize collective communication features for scale-up and scale-out networks; Profile and tune distributed AI workloads (LLM, diffusion, multimodal); Drive features end-to-end from design to delivery; Develop benchmarking and performance estimation tools; Collaborate with GPU library, runtime, and open-source teams; Contribute to hardware-software co-design discussions; Mentor junior engineers
Seniority
Sr. Staff, hands-on IC with mentorship
