HPC System Architecture

About the role

Job Details:

Job Description:

The HPC System Architect is the senior technical lead responsible for defining the next generation of Intel's leading-edge high-performance computing systems. Reporting directly to Nic Dube, Corporate Vice President and General Manager of Data Center Systems and Solutions, this position owns forward-looking HPC system architecture from concept through committed definition.

You will work directly with Intel's CPU and GPU teams to architect the systems that follow today's products - shaping silicon requirements, and defining system-level architecture for the large-scale, scale-out clusters that characterize production HPC. The role spans full-system definition: compute node design, memory and I/O balance, scale-out fabric and topology, and the software stack that determines delivered application performance at scale.

You will engage directly with the customers that set the leading edge of HPC - DOE national laboratories and their exascale and post-exascale programs, financial services institutions, oil and gas operators, and major academic and government research computing centers - translating their workload and facility requirements into system architectures Intel can build, and Intel's roadmap into system concepts they can commit to.

Responsibilities

- Own system-level architecture for next-generation HPC clusters: node definition, CPU and GPU configuration, memory bandwidth-to-capacity balance, I/O and parallel storage architecture, and scale-out fabric and topology.

- Define scale-out interconnect architecture - fabric selection, network topology, bisection bandwidth and blocking ratios, collective and communication performance, and resilience at thousands-of-node scale.

- Partner with the CPU and GPU teams to define product requirements for future HPC systems, bringing system-level and workload evidence into silicon definition while requirements remain open.

- Build performance models that project delivered application throughput at scale on unreleased silicon and use them to drive configuration and roadmap decisions.

- Lead architecture co-design and benchmark definition with leadership HPC customers, including acceptance criteria for standard and customer-proprietary application suites.

- Drive HPC requirements into Intel's CPU, GPU, platform, and software roadmaps across DCG and the product and design organizations.

- Evaluate emerging technologies for next-generation HPC systems, including advanced memory, Ultra Ethernet and other scale-out fabric standards, optical interconnect, and heterogeneous accelerator integration.

- Produce reference architectures and system specifications that platform teams, OEM and ODM partners, and system integrators can build to.

- Grow and lead HPC architecture capability within the team, providing technical direction and mentorship to architects and engineers.

Successful Candidate Profile