Principal / Senior GPU SW Performance Engineer — Post‑Training
Core
Drive performance of post-training workloads (finetuning, RL) on AMD Instinct GPUs by optimizing kernels, distributed training, and framework integrations for fast, stable pipelines on ROCm.
Role type
Senior IC GPU software performance engineer (AI training)
Builds
High-performance training pipelines and reproducible workflows for AI/ML teams
Domain
AI/Deep Learning, High-Performance Computing (HPC), GPU Systems
Deliverable
production ML models
Required skills
GPU performance engineering, distributed training optimization, kernel development, profiling and bottleneck resolution, multi-GPU/multi-node communication patterns, PyTorch (torch.distributed, FSDP/ZeRO), Python, C++, ROCm/HIP/Triton
Preferred skills
SFT, LoRA, RL-based training at scale, upstreaming changes to frameworks, distributed systems and collective communication libraries
Technologies
ROCm, HIP, Triton, PyTorch, C++, Python
Responsibilities
Lead performance for finetuning and RL training solutions; Improve throughput, memory efficiency, and stability across data, model, and optimizer steps; Optimize multi-GPU/multi-node training and communication patterns; Contribute efficient kernels/ops and targeted graph-level optimizations; Profile, diagnose, and resolve bottlenecks using standard tooling; Ship reproducible pipelines and documentation; Collaborate with framework, compiler, and model teams to land durable improvements.
Seniority
Senior, hands-on IC
