The role involves optimizing and developing deep learning frameworks specifically for AMD GPUs. As a core team member, you will enhance GPU kernel performance and accelerate deep learning models, while also enabling reinforcement learning training and state-of-the-art large language model inference at scale. Collaboration with internal GPU software teams and open-source communities is essential to integrate and optimize cutting-edge compiler technologies. The position requires a skilled engineer with strong technical expertise in GPGPU C++ and related technologies.

• Strong technical and analytical expertise in GPGPU C++, Triton, TileLang or DSL development within Linux environments.
• Proficient in C++ and/or Python with demonstrated ability to code, debug, profile, and optimize performance-critical code.
• Hands-on experience with SGLang or similar LLM inference frameworks is highly preferred.
• Background in compiler design or familiarity with technologies like LLVM, MLIR, or ROCm is a plus.
• Experience running and scaling workloads on large-scale, heterogeneous clusters using distributed training or inference strategies.

• Enhance performance of frameworks like TensorFlow, PyTorch, and SGLang on AMD GPUs via upstream contributions in open-source repositories.
• Profile, analyze, code change and tune large-scale training and inference models for optimal performance on AMD hardware.
• Design, implement, and optimize high-performance GPU kernels using HIP, Triton, TileLang or other DSLs for AI operator efficiency.
• Work closely with internal compiler and GPU math library teams to integrate, optimize and align kernel-level optimizations with full-stack performance goals.
• Support optimization, feature development, and scaling of the SGLang framework across AMD GPU platforms for LLM, multimodal serving and RL-training.

• AMD provides a competitive 'Total Rewards' package that focuses on financial growth, health, and work-life balance.