Senior Memory & Storage Architect for Large-Scale LLMs

NVIDIA Corporation; Santa Clara (CA); Presencial; USD 299.000 - 404.000; Jornada completa

Descripción de la vacante

NVIDIA Corporation seeks a Principal Systems Engineer to define the memory management vision for large-scale LLM and storage systems. You will design a unified memory layer spanning GPU memory, host memory, and remote storage to power disaggregated AI clusters.

You will collaborate with GPU architecture, networking, and platform teams to leverage GPUDirect, RDMA, NVLink, and related technologies, mentor engineers, and drive practical, production-grade solutions across multi-node environments.

Formación

• 15+ years of experience building large-scale distributed systems, high-performance storage, or ML systems infrastructure in C/C++ and Python.
• Deep understanding of memory hierarchies (GPU HBM, host DRAM, SSD, and remote/object storage).
• Distributed caching or key-value systems with low latency & high concurrency.
• Hands-on experience with networked I/O and RDMA/NVMe-oF/NVLink-style technologies.
• Strong profiling and optimization across CPU, GPU, memory, and network.
• Excellent communication skills and cross-functional leadership.

Responsabilidades

• Design and evolve a unified memory layer spanning GPU memory, pinned host memory, RDMA-accessible memory, SSD tiers, and cloud storage.
• Architect and implement deep integrations with LLM serving engines (e.g., vLLM, SGLang, TensorRT-LLM) for KV-cache offload and sharing.
• Co-design interfaces for disaggregated prefill, peer-to-peer KV-cache sharing, and multi-tier KV-cache storage.
• Partner with GPU, networking and platform teams to exploit GPUDirect, RDMA, NVLink for low-latency KV-cache access.
• Mentor engineers and represent the team in reviews and external forums.

Herramientas

RDMA/NVMe-oF/NVLink

C/C++

Python

Descripción del empleo

NVIDIA Corporation seeks a Principal Systems Engineer to define the memory management vision for large-scale LLM and storage systems. You will design a unified memory layer spanning GPU memory, host memory, and remote storage to power disaggregated AI clusters.

You will collaborate with GPU architecture, networking, and platform teams to leverage GPUDirect, RDMA, NVLink, and related technologies, mentor engineers, and drive practical, production-grade solutions across multi-node environments.