Member of Technical Staff - GPU Infrastructure

Remote · United States | San FranciscoRemoteFull-time$150k–$300k2026-11-08 → 2026-11-11

HybridVisa sponsorship offered

Core

Design and deploy high-performance GPU cluster architectures for frontier AI workloads, serving as the primary technical escalation point for customer infrastructure issues.

Role type

Senior IC GPU Infrastructure Engineer (Customer-Facing)

Builds

GPU clusters (100–10,000+ GPUs) for LLM training, inference, and HPC workloads

Domain

AI Infrastructure / High-Performance Computing

Deliverable

infrastructure

Required skills

GPU cluster architecture, SLURM, Kubernetes, InfiniBand configuration, NVIDIA CUDA ecosystem, Linux kernel tuning, Python, Bash, systems programming

Preferred skills

1000+ GPU deployments, NVIDIA DGX/HGX/SuperPOD certification, distributed training frameworks (PyTorch FSDP, DeepSpeed), AMD MI300/Intel Gaudi experience

Technologies

SLURM, Kubernetes, InfiniBand, RoCE, NVLink, Lustre, BeeGFS, GPFS, CUDA, Docker, Containerd, Ansible, Terraform

Responsibilities

Partner with clients to design optimal GPU cluster architectures; Deploy and configure orchestration systems and high-performance networking; Serve as primary technical escalation point for complex hardware, driver, and software issues; Optimize GPU utilization, memory management, and inter-node communication.

Seniority

Senior, hands-on IC

About Prime Intellect

Seed50 Employees

Prime Intellect builds the infrastructure that frontier AI labs build internally, and makes it available to everyone. Our platform, Lab, unifies environments, evaluations, sandboxes, and high-performance training into a single full-stack system for post-training at frontier scale, from RL and SFT to tool use, agent workflows, and deployment. We validate everything by using it ourselves, training open state-of-the-art models on the same stack we put in your hands. We're looking for people who want to build at the intersection of frontier research and real infrastructure.