Sr. Software Engineer - AI Triton Communication
Core
Building native distributed execution and communication capabilities for the Triton compiler and runtime stack to enable efficient large-scale AI training and inference on AMD Instinct Accelerators.
Role type
Senior IC software engineer (GPU compiler/runtime/distributed systems)
Builds
Native distributed execution and communication primitives for Triton on AMD CDNA and next-gen GPUs
Domain
High-performance computing, AI infrastructure, GPU architecture
Deliverable
production ML models
Required skills
GPU architecture, compiler development, distributed systems, performance engineering, Triton compiler/runtime, MLIR/LLVM internals, GPU programming (HIP/CUDA/Triton), multi-GPU optimization, ROCm stack
Preferred skills
Triton compiler/runtime experience, GPU runtime/communication stacks (XGMI, NVLink, InfiniBand), distributed libraries (RCCL, NCCL, NVSHMEM, MPI), profiling/debugging across stack, open-source contribution
Technologies
Triton, ROCm, HIP, CUDA, MLIR, LLVM, XGMI, NVLink, PCIe, InfiniBand, RCCL, NCCL, NVSHMEM, rocSHMEM, MPI
Responsibilities
Design and develop native distributed communication and execution capabilities within the Triton AMDGPU backend; Design and implement Triton compiler and runtime mechanisms for native GPU-initiated communication; Drive performance optimization across compute and communication; Develop and optimize distributed Triton kernels and execution models; Analyze, profile and debug complex cross-stack issues; Collaborate with GPU architecture, compiler, runtime, and performance teams; Contribute to open-source Triton and ROCm distributed ecosystem
Seniority
Senior, hands-on IC
