18 hours ago
Responsibilities
- Lead performance modeling and optimization for multi-trillion-parameter LLM training and inference, including dense, mixture-of-experts, and multimodal models.
- Model and optimize tensor, pipeline, context, expert, and data parallelization strategies.
- Architect memory-efficient training systems using structured pruning, quantization, continuous batching, chunked prefill, and speculative decoding.
- Incorporate and extend advanced models including GPT-4, DeepSeek-R1, and multimodal architectures.
- Collaborate with internal and external stakeholders and ML researchers to communicate results and iterate on system designs.
- Define GPU architecture specifications and distributed ML system best practices for scaling from single-GPU to thousand-GPU deployments.
Requirements
- Extensive senior-level experience optimizing large-scale ML systems and GPU architectures.
- Deep expertise in CUDA programming, GPU memory hierarchies, and hardware-specific optimization.
- Proven experience architecting distributed training systems at large scale.
- Expert knowledge of transformer architectures, attention mechanisms, and model-parallelism techniques.
- Bachelor’s, master’s, or PhD in Computer Science or Engineering, or equivalent industry experience.
- Strong software engineering, technical leadership, cross-functional collaboration, and communication skills.
- Preferred experience with PyTorch, CUDA, TensorRT, OpenAI Triton, Ray, Megatron-LM, Nsight Compute, nvprof, PyTorch Profiler, KV-cache optimization, Flash Attention, mixture-of-experts systems, InfiniBand, RDMA, and NVLink.
Benefits
- AMD benefits are offered; details are provided in AMD’s benefits overview.
- This role requires working five days per week in an AMD office location.
Tech Stack
Categories
About AMD
AMD designs and sells CPUs, GPUs, and adaptive/embedded computing products for PCs, data centers, gaming, and edge devices. Its portfolio includes Ryzen and EPYC processors, Radeon and Instinct graphics, and adaptive SoCs from its Xilinx acquisition, sold to OEMs, cloud providers, and device makers. Founded in 1969 and headquartered in Santa Clara, it is a public company on NASDAQ and supplies semi-custom chips for major game consoles.