Compute Infrastructure Engineer for Frontier AI

OpenAI; California (MO); On-site; USD 180,000 - 260,000; Full time

Job summary

OpenAI is seeking engineers to build and operate the compute platform that powers its research and products. You will work on large-scale, heterogeneous clusters spanning accelerators, CPUs, networks, and storage, with a focus on performance, reliability, and developer experience.

You may specialize near hardware or near users, contributing to CaaS, agent infrastructure, control/data planes, and tooling to make compute systems faster and easier to use.

Qualifications

• Experience building, operating, or improving production infrastructure systems.
• Strong background in distributed systems, networking, or HPC environments.
• Familiarity with Kubernetes, NCCL, RDMA, and related tooling.

Responsibilities

• Build and deeply optimize reliable system software for large-scale compute systems.
• Design and operate infrastructure across accelerators, CPUs, NICs, switches, storage, and scheduling.
• Profile, benchmark, and optimize training workloads across compute and networking bottlenecks.
• Create hardware-aware automation for provisioning, firmware updates, and incident response.
• Build CaaS, agent infrastructure, profiling, observability, benchmarking, and platform tools.
• Turn operational lessons into better systems and clearer ownership boundaries across teams.
• Collaborate across research, engineering, security, networking, hardware, and data center teams.

Skills

Strong software engineering

Distributed systems

Kubernetes

GPU infrastructure

CaaS

Observability

Profiling & benchmarking

Tools

NCCL

RDMA

GPU hardware tooling

Job description

OpenAI is seeking engineers to build and operate the compute platform that powers its research and products. You will work on large-scale, heterogeneous clusters spanning accelerators, CPUs, networks, and storage, with a focus on performance, reliability, and developer experience.

You may specialize near hardware or near users, contributing to CaaS, agent infrastructure, control/data planes, and tooling to make compute systems faster and easier to use.