Job SummaryWe are seeking a highly accomplished experienced GPU Architect to define the next generation of AI accelerators and multi-GPU cluster architecture. As the demand for trillion-parameter LLM training and high-throughput localized inference accelerates, the role of GPU architecture has never been more critical. In this role, you will lead the technology characterization, reliability, and interconnect performance strategies that ensure our compute fabrics scale flawlessly. You will collaborate deeply across hardware, firmware, and AI silicon teams to build GPU infrastructure capable of pushing the absolute limits of parallel processing and hardware efficiency.
Responsibilities and Duties- Hardware-Software Co-Design: Collaborate with software engineering to ensure the AI compute and Rack level hardware architectures fundamentally accelerate lower-level ML frameworks and localized inference engines (e.g., vLLM, Ollama, TensorRT).
- Performance Modeling: Build and analyze cycle-accurate simulators and analytical models to identify bottlenecks, forecast workload performance, and guide architectural trade-offs.
- Influence long-term silicon architecture roadmaps with our GPU SoC teams. Mentor engineering teams and drive strict engineering standards from feasibility to tape-out and post-silicon validation.
- Reliability: As a Platform level GPU architect, the role requires the candidate to have extensive knowledge in Reliability and Quality including but not limited to the ability to calculate MTBF, FIT rates, IEFR, IFR, and lifecycle bath-tub curves to understand repair rates, SLAs, uptime curves.
- NPI Manufacturing: The role requires a deep knowledge with manufacturing processes to detect and correct any inadequate manufacturing frameworks that can impact the overall quality of the products we deploy in our Datacenters.
Candidate ProfileEssential:- Experience: 10+ years of deep experience in GPUs, AI accelerators, or highly parallel computer systems in areas of qualification, manufacturing, and programming.
- Microarchitecture Expertise: Understanding of SIMD/SIMT execution models, instruction scheduling, and hardware acceleration for machine learning algorithms.
- Manufacturing: Deep knowledge of advanced manufacturing techniques for build of AI compute units and Rack level L11 liquid cooled solutions.
- Systems Interconnects: Extensive hands-on experience characterizing data pathways across RDMA environments, and hardware clustering protocols.
- Programming & Tooling: Proficiency in C++, Python, or similar languages for performance modeling, GPU technology characterization, and workload profiling.
- Analytical Rigor: Exceptional ability to characterize complex AI mathematical operations into efficient hardware implementations.
- Education: BS or MS or equivalent experience in Computer Engineering or Electrical Engineering.
Desirable- Specific Topology Experience: Direct experience qualifying Rack-scale GPU designs including but not limited to NPI manufacturing, testing, quality and reliability calculations.
