Senior Engineer, Compute Fabric & Network Architecture, Project Sunrise

BLUE ORIGIN•Denver, CO

4d•$133,500 - $203,263

About The Position

At Blue Origin, we envision millions of people living and working in space for the benefit of Earth. We’re working to develop reusable, safe, and low-cost space vehicles and systems within a culture of safety, collaboration, and inclusion. Join our team of problem solvers as we add new chapters to the history of spaceflight! This role is part of Advanced Concepts and Enterprise Engineering (ACE), supporting Blue Origin’s mission of millions of people living and working in space for the benefit of Earth. The team fosters innovation and drives engineering workflows of the future, shared solutions and standards, simplicity and lower costs, and manufacturable design. Our team is developing next-generation spacecraft that push the boundaries of performance and capability in orbit. We are building a large-scale orbital data center, and we are looking for a hardware-leaning systems engineer to be the architect and builder of its nervous system. This is not an operator role; you will be the owner responsible for making the foundational hardware decisions for our high-speed interconnect fabric and owning its performance from the physical layer all the way up to the software stack. You will solve novel, first-principles challenges, such as ensuring fabric resiliency in a radiation environment, and your decisions will be critical to the success of the entire program.

Requirements

Bachelor of Science in Computer Science, Electrical Engineering, or a related field.
8+ years of experience in HPC networking, datacenter fabric design, or GPU cluster engineering.
Demonstrated expertise in designing and deploying large-scale accelerator clusters using high-speed, direct GPU-to-GPU interconnects and high-radix fabric switches.
Deep, hands-on experience with network topology at scale (e.g., large supercomputer clusters or hyperscale infrastructure) and associated fabric management tools.
Proficiency with collective communication libraries.
Must be a U.S. citizen or national, U.S. permanent resident (current Green Card holder), or lawfully admitted into the U.S. as a refugee or granted asylum.

Nice To Haves

Master’s or Ph.D. in a relevant discipline.
10-15+ years of experience in a senior networking or systems design role.
Extensive experience with InfiniBand fabrics.
Intellectual curiosity and a first-principles approach to problem-solving, particularly for novel challenges like the effect of radiation on network switch silicon (prior space experience is not required).
Experience at a major cloud provider (e.g., Meta, Google, Microsoft), HPC vendor, or National Laboratory (e.g., ORNL, ANL, LLNL) focused on large-scale GPU networking.
Passion for and understanding of AI and Machine Learning applications in engineering, with a demonstrated ability to learn and integrate new technologies.

Responsibilities

Architect the hierarchical GPU interconnect topology for a large-scale accelerator cluster, defining pod architecture, fabric switch placement, and inter-pod fabric selection.
Following hardware selection, own the configuration, bring-up, and operation of fabric management software, defining topology files and partitioning policies for multi-tenant workloads.
Lead the hardware qualification of fabric switches and other network components for the space environment, developing novel strategies to mitigate radiation effects on switch state and routing tables.
Take ownership of the collective communications performance across the fabric, tuning libraries (e.g., NCCL, RCCL) and defining the bandwidth/latency budgets required for high-performance model-parallel inference.
Collaborate deeply with software teams to co-design the fabric topology with model parallelism strategies, ensuring optimal performance for large-scale AI models.
Interface with platform software teams to manage the fabric software bring-up and integration process.