Software Development · San Francisco, CA

Full-time

Software Engineer, GPU Infrastructure - HPC

OpenAI · San Francisco, CA · Posted Feb 5, 2026

This posting comes from OpenAI. Confirm the details on their site before applying.

Overview

The Fleet team at OpenAI supports the computing environment that powers our cutting-edge research and product development. We oversee large-scale systems that span data centers, GPUs, networking, and more, ensuring high availability, performance, and efficiency. Our work enables OpenAI’s models to operate seamlessly at scale, supporting both internal research and external products like ChatGPT.

Responsibilities

  • Build and maintain automation systems for provisioning and managing server fleets.
  • Develop tools to monitor server health, performance, and lifecycle events.
  • Collaborate with clusters, networking, and infrastructure teams.
  • Partner with external operators to ensure a high level of quality.
  • Identify and fix performance bottlenecks and inefficiencies.
  • Continuously improve automation to reduce manual work.

Requirements

  • Experience managing large-scale server environments.
  • A balance of strengths in building and operationalizing.
  • Proficiency in Python, Go, or similar languages.
  • Strong Linux, networking, and server hardware knowledge.
  • Comfort digging into noisy data with SQL, PromQL, and Pandas or any other tool.

Preferred qualifications

  • Experience with low level details of hardware components, protocols, and associated Linux tooling (e.g., PCIe, Infiniband, networking, power management, kernel perf tuning)
  • Knowledge of hardware management protocols (e.g., IPMI, Redfish).
  • High-performance computing (HPC) or distributed systems experience.
  • Prior experience developing, managing, or designing hardware.
  • Familiarity with monitoring tools (e.g., Prometheus, Grafana).

Benefits

  • Bonus

Great fits start here.

Whether you're looking for your next role or your next team member, give people and possibilities a place to meet.