Sr. Software Engineer - AI Triton Kernels
Core
Authoring and optimizing high-performance Triton/Gluon kernels for large-scale LLM and multimodal AI workloads on AMD Instinct GPUs.
Role type
Senior IC software engineer (AI kernel development & performance engineering)
Builds
Production Triton kernels for vLLM and SGLang frameworks
Domain
AI/ML infrastructure, GPU computing, compiler backends
Deliverable
production ML models
Required skills
SIMT programming, GPU architecture, parallel algorithms, performance engineering, Triton, Triton IR lowering, ROCm, LLVM AMDGPU stack, memory hierarchy optimization, wavefront execution, vectorization, MFMA utilization, hardware profiling, microbenchmarking
Preferred skills
Open-source contributions (Triton, Torch, vLLM, SGLang, IREE, MLIR, LLVM, ROCm), deep understanding of modern GPU architectures
Technologies
Triton, Gluon, PyTorch, vLLM, SGLang, ROCm, LLVM, MLIR, CDNA, HBM, LDS, L2
Responsibilities
Design and optimize matmul, attention, MoE, and fused transformer kernels; Partner with compiler engineers to maintain the Triton AMD backend; Drive kernel-level optimizations across memory hierarchy and instruction scheduling; Perform profiling and root-cause analysis of performance bottlenecks; Debug correctness issues across runtimes and backends; Contribute to open-source ecosystems
