Cette offre d'emploi n'est plus disponible
À propos
Ensure you read the information regarding this opportunity thoroughly before making an application.
Key Responsibilities Design and implement compiler infrastructure for GPU-accelerated workloads, including lowering pipelines and code generation backends. Develop and optimize MLIR-based compilation passes targeting Oxmiq hardware IP and GPU architectures. Build and maintain LLVM backend components, including instruction selection, register allocation, and target-specific optimizations. Work with architecture and hardware teams to translate performance requirements into efficient code generation strategies. Optimize compiler output for throughput, latency, and memory efficiency across GPU workloads. Collaborate with kernel engineers to support Triton-based programming models and custom op lowering. Participate in design reviews, code reviews, and architecture discussions. Support performance profiling, benchmarking, and root cause analysis for compiler-generated code. Required Qualifications 5+ years of compiler engineering experience with a focus on GPU or hardware accelerator backends. Strong expertise in C++ and LLVM, including familiarity with the LLVM IR and pass infrastructure. Hands-on experience with MLIR, including dialect design, conversion passes, and progressive lowering. Proven experience with GPU programming models, including CUDA. Solid understanding of code generation concepts including instruction scheduling, vectorization, and memory hierarchy optimization. Familiarity with GPU architecture fundamentals including thread execution models, memory systems, and occupancy constraints. Experience working with Triton or similar high-level GPU kernel frameworks. Strong debugging and performance analysis skills. Hands-on experience with Claude Code or equivalent AI-assisted development workflows. Preferred Qualifications Experience developing compiler backends for custom AI accelerators, NPUs, or GPU IP. Familiarity with kernel fusion, operator tiling, and auto-tuning frameworks. Exposure to quantization, mixed-precision, or sparsity-aware compilation techniques. Experience with performance modeling and roofline analysis for GPU workloads. Contributions to open-source compiler projects such as LLVM, MLIR, or Triton. Scripting experience in Python for build automation, testing, or performance benchmarking. xywuqvp Education BS/MS/PhD in Computer Science, Computer Engineering, Electrical Engineering, or a related field.
Compétences linguistiques
- English
Avis aux utilisateurs
Cette offre a été publiée par l’un de nos partenaires. Vous pouvez consulter l’offre originale ici.