About the job
AMD is seeking a Fellow of EPYC AI Product Architecture to lead the definition of next-generation CPU and platform innovations tailored for AI workloads. You will be a key technical leader within the EPYC AI Product organization, shaping AMD’s AI platform strategy across silicon, systems, and software. This role sits at the intersection of architecture, product definition, customer engagement, and business impact, driving differentiated solutions across cloud, enterprise, and hyperscale deployments.
Responsibilities
Lead architecture definition for AMD EPYC CPU and server platforms optimized for AI training and inference
Engage with hyperscalers, OEMs, and AI ISVs to align platform features with evolving workload needs
Evaluate and drive new CPU and platform features for deep learning models, including generative AI, vision, and recommender systems
Analyze performance bottlenecks using architecture simulation and hardware instrumentation; propose workload-driven improvements
Drive architectural trade-off analyses across compute, memory, I/O, and network subsystems
Build and refine performance models, automation tools, and workload testbeds for end-to-end analysis
Project and compare performance vs TCO tradeoffs under different system and silicon configurations
Shape AMD’s platform strategy for heterogeneous compute, working closely with GPU and AI accelerator teams
Represent AMD in industry forums, customer briefings, analyst interactions, and press engagements
Qualifications
Minimum
No minimum qualifications listed.
Preferred
Several years in high-performance CPU, server, or AI platform architecture, ideally with customer-facing responsibilities
Expertise in AI system deployments at scale (cloud, enterprise, HPC, or edge)
Demonstrated thought leadership in Generative AI (LLMs), vision, or recommender systems
Hands-on experience with performance tools, roofline models, and system simulation
Familiarity with AI compilers, quantization flows (QAT/PTQ), and workload optimization techniques
Proficient in deep learning frameworks such as PyTorch, TensorFlow, and inference runtimes like ONNX Runtime or TensorRT
Understanding of model deployment pipelines, sparsity techniques, advanced numeric formats, and mixed precision
Optional: CUDA programming or Hugging Face pipelines
Track record of cross-functional leadership and working in fast-paced, ambiguous environments