Software Engineer, GPU AI ML

What you'll do

• Optimize performance across the entire AI software stack from low-level GPU kernels to large-scale distributed systems.
• Improve training, post-training, and inference performance for LLMs and autonomous AI workloads on AMD GPUs.
• Develop and optimize high-performance C++/HIP/CUDA codebases within the ROCm ecosystem.
• Identify and resolve performance bottlenecks in multi-GPU, distributed settings using specialized profiling tools.
• Implement advanced techniques such as quantization, speculative decoding, and MoE for large-scale inference pipelines.
• Partner on hardware-software co-design to influence GPU architecture and new accelerator development.
• Establish best practices and mentor others to scale AMD's AI software strategy.

What we're looking for

• Expert-level proficiency in modern C++ and designing large, performance-critical systems.
• Strong knowledge of GPU architecture, memory hierarchy, and kernel optimization using HIP or CUDA.
• Experience with large-scale C++/HIP/CUDA codebases including ROCm, the CUDA ecosystem, and ML framework cores like PyTorch, TensorFlow, or JAX.
• Proficiency in diagnosing bottlenecks using profilers in multi-GPU, distributed settings.
• Deep understanding of transformer architectures, attention mechanisms, and the full model lifecycle.
• Hands-on experience in alignment and post-training techniques such as SFT, RLHF, and GRPO.
• Knowledge of LLM trends including MoE, quantization, speculative decoding, and agentic systems.
• Bachelor's degree in Computer Science, Computer Engineering, Electrical Engineering, or equivalent experience; Master's or PhD preferred.