Principal Architect, AI Networking

Description

The company's applied research team, part of its networking systems and software architecture group, works on hard infrastructure problems in AI. The team builds systems software that moves data between GPUs, nodes and storage fast enough for current AI workloads, spanning transport-level tuning, joint hardware and software design, and communication frameworks that connect directly to live AI stacks.

The team's scope also extends into new areas, including interconnects for quantum computing.

This principal role sets the research direction and architecture for how the company's AI systems communicate at scale across GPUs, DPUs, NICs and mixed storage. You define projects from the ground up, publish original work and turn research results into production software used across the industry.

Responsibilities

• Define the long-term technical vision for distributed AI communication, including GPU-to-GPU, GPU-to-storage and cross-node data movement.
• Run original research and build prototypes of next-generation networking over RDMA, NVLink and GPUDirect.
• Co-optimize hardware and software across the GPU, DPU, NIC and network switch.
• Find the root causes of bottlenecks in communication runtimes for large AI workloads, such as KV cache transfer, disaggregated prefill and decode, and model parallelism.
• Bring networking capabilities into AI serving stacks such as vLLM, SGLang and TensorRT-LLM.
• Publish findings, represent the company at industry forums and standards bodies, and mentor senior engineers across the organization.

Requirements

• 15+ years in systems software, networking or both, with deep high-performance networking knowledge (InfiniBand, RoCE, RDMA, NVLink), experience with communication libraries such as NIXL, NCCL, UCX, MPI or NVSHMEM, and work on GPU-accelerated systems.
• A track record of taking complex, cross-team technical work from research concept to production.
• MS, PhD or equivalent experience in computer science, computer engineering, electrical engineering or a related field.
• Strong grasp of computer architecture, memory hierarchies, DMA engines and OS-level networking.
• Working understanding of ML systems, such as transformer architectures, KV cache mechanics, model parallelism, or distributed training and inference patterns.
• Proficiency in languages such as C, C++, Rust and Python.
• Ways to stand out: knowledge of vLLM, SGLang and TensorRT-LLM inference frameworks and their communication needs.
• Ways to stand out: CUDA programming and expertise in the company's GPU architecture.
• Ways to stand out: a proven record of influencing product strategy and technical roadmap at a senior level.
• Ways to stand out: significant open-source contributions.

Conditions

Base salary is set by location, experience and other objective job-related factors, with an expected range of 272,000 USD to 431,250 USD.

This is an existing vacancy, and applications are accepted at least until November 3, 2026.

The company uses AI tools in its recruiting processes.

The company is committed to an inclusive workplace, is an equal opportunity employer and does not discriminate on any legally protected characteristic.

Benefits

• Competitive salary and a comprehensive benefits package.