✦ Luna Orbit — Software Engineering

Principle Triton Compiler Engineer

at Advanced Micro Devices

📍 San Jose, California, United States Hybrid Posted March 14, 2026
Type Full-Time
Experience lead
Exp. Years Not specified
Education Not specified
Category Software Engineering

This role focuses on advancing the Triton compiler stack for AMD GPUs, including building compiler infrastructure, optimizing kernels, and collaborating with hardware teams to improve performance and usability.

  • Develop Triton compiler backend; Optimize GPU kernels; Improve memory and execution efficiency; Collaborate with architecture teams; Enhance compiler infrastructure

The technical environment includes compiler backends like LLVM and MLIR, AMD's ROCm platform, GPU architecture, and performance profiling tools, aimed at optimizing Triton kernels for AMD hardware.

The ideal candidate is a lead compiler engineer with deep expertise in GPU compiler technologies, performance optimization, and GPU architecture. They should have experience working with LLVM, MLIR, and AMD GPU backends, and be capable of developing high-performance kernels and compiler infrastructure.

Compiler technologiesGPU architecturePerformance engineeringIR transformationsLLVMMLIRCode generationKernel tuning
Memory hierarchy optimizationWave-level executionOccupancy optimizationInstruction schedulingKernel profilingROCmAMD GPU backend
LLVMMLIRROCmGitLinux
Compiler technologiesGPU architecturePerformance engineeringIR transformationsLLVMMLIRCode generationKernel tuningMemory hierarchyWave-level executionOccupancyInstruction schedulingKernel profilingTritonROCmAMD GPU backend
Compiler TechnologiesGPU ArchitecturePerformance EngineeringIR TransformationsLLVMMLIRCode GenerationKernel TuningMemory HierarchyWave-level ExecutionOccupancyInstruction SchedulingKernel Performance ProfilingTritonROCmLLVM AMDGPU Backend
CollaborationProblem-SolvingTechnical CommunicationAnalytical ThinkingCuriosityHands-on Approach
Industry Semiconductors & Electronics
Job Function Enhancing Triton compiler and kernel performance for AMD GPUs
Role Subtype Software Engineering
Tech Domains LLVM, MLIR, ROCm, GPU, Compiler Technologies, Performance Engineering
compiler technologiesgpu architectureperformance engineeringIR transformationsLLVMMLIRcode generationkernel tuningmemory hierarchywave-level executionoccupancyinstruction schedulingkernel profilingtritonrocmamdgpu backend

Lack of experience with compiler technologies, No background in GPU architecture or performance engineering, Unable to work in a hybrid environment, No experience with AMD GPUs

Apply for this Position →

Get matched to jobs like this

Luna finds roles that fit your skills and career goals — no endless scrolling required.

Create a Free Profile