Principal AI Performance Engineer

What you'll do

• Drive end-to-end performance optimization on AMD GPUs for leading AI models.
• Profile and resolve complex cross-stack bottlenecks in GPU kernels and frameworks.
• Diagnose kernel-level issues using profiling tools to enhance model performance.
• Lead customer engagements by presenting technical findings and optimizations.
• Develop custom kernels within serving frameworks to improve dispatch efficiency.
• Optimize multi-node distributed inference for better communication-compute overlap.
• Define and refine performance optimization methodologies for the broader team.

What we're looking for

• 7+ years of software development experience in GPU computing, AI systems, or high-performance computing.
• Deep hands-on experience with AI serving frameworks and their internals, including vLLM, SGLang, TensorRT-LLM.
• Strong background in end-to-end workload profiling and bottleneck diagnosis from user request to GPU kernel.
• Expertise in GPU kernel performance characteristics such as occupancy, memory coalescing, cache utilization, and instruction-level bottlenecks.
• Experience with custom kernel development or integration using HIP, CUDA, Triton, CK, or similar technologies.
• Understanding of multi-GPU and multi-node distributed systems, including scale-up and scale-out topologies, RDMA, and communication-compute overlap.
• Fluent in AI-assisted development, leveraging AI agents and tools daily to accelerate workflows.