Architect communication libraries and collective algorithms across NVLink, Spectrum-X, and Quantum-X interconnects.
Originally posted October 30, 2026 by the employer. Relisted by the employer on October 31, 2026.
About the role
NVIDIA is seeking a Principal Deep Learning Communication Architect to define the long-term technical roadmap for communication libraries across compute platforms. The role focuses on scaling AI and HPC libraries across heterogeneous interconnects such as NVLink, Spectrum-X, and Quantum-X for trillion-parameter models.
What you'll do
- Define the technical roadmap for communication libraries to enable model scaling across clusters of hundreds of thousands of nodes.
- Lead the development of communication primitives and collective algorithms optimized for NVLink, Spectrum-X (Ethernet), and Quantum-X (InfiniBand).
- Partner with application developers to architect and implement specialized communication primitives across libraries including NCCL, NIXL, NVSHMEM, UCC, and UCX.
- Collaborate with silicon architects and software engineers on hardware/software co-design to influence hardware specifications for networking.
- Develop high-fidelity analytical models and simulators to predict system behavior under emerging workloads.
What you'll need
- Ph.D. or M.S. in Computer Science, Electrical Engineering, or a related field (or equivalent experience), with 12+ years of industry experience in high-performance computing (HPC) or distributed deep learning.
- Deep understanding of 3D parallelism (Data, Tensor, Pipeline) and advanced strategies including Context Parallelism, Expert Parallelism, and Zero Redundancy Optimizer (ZeRO) variants.
- Deep technical proficiency with NCCL, UCX, UCC, NVSHMEM, or MPI, alongside experience with RDMA, RoCE, and low-level InfiniBand verbs.
- Advanced knowledge of high-throughput inference engines and schedulers, specifically TensorRT-LLM, vLLM, SGLang, and NVIDIA Dynamo.
- Expert knowledge of the NVIDIA GPU memory hierarchy (HBM3e/HBM4, L2 cache) and CUDA programming models.
Nice to have
- Hands-on experience developing within Megatron-Core, DeepSpeed, or JAX/XLA, with an understanding of runtime interactions with low-level communication runtimes.
- Significant upstream contributions to major open-source projects such as PyTorch Distributed, KServe, or Ray.
- Track record of deploying and optimizing models on NVIDIA platforms or rack-scale systems.
- Portfolio of patents or papers in systems/architecture venues such as ISCA, ASPLOS, NeurIPS, or SC.
Skills: trillion-parameter AI, hardware/software co-design, HPC, distributed deep learning, GPU architecture, NCCL, RDMA
More semiconductor jobs: NVIDIA jobs · AI chip design jobs
Tracked in our index since 15 Apr 2026. This is our observation date, not the employer posting date. Specialty: AI/ML Hardware Engineering (AI ML Hardware). 477 active roles in this specialty at 76 employers in our index, as of 19 Nov 2026.
How we counted: postings read from employer career sites. How we count