Role Overview

The Fleet team at OpenAI powers the computing backbone for our groundbreaking AI research and products. We manage exascale infrastructure spanning thousands of GPUs across data centers worldwide, ensuring 99.99% availability for mission-critical AI workloads. Your work directly enables researchers to train next-generation models while maintaining safety and reliability at unprecedented scale.

In this role, you'll bridge low-level hardware systems with high-level orchestration, creating tools that integrate bare-metal servers into unified clusters. You'll leverage cutting-edge LLMs to automate vendor operations, optimize scheduling algorithms, and eliminate operational toil. This position requires deep systems expertise and passion for solving infrastructure challenges that don't exist anywhere else.

Based in San Francisco with a hybrid model (3 days/week in office), this role offers relocation support and exposure to the most advanced AI infrastructure on the planet. You'll collaborate with world-class researchers, hardware engineers, and infrastructure experts pushing the boundaries of what's possible in AI compute.

Key Responsibilities

Software Engineers on the Fleet team tackle complex challenges across the entire infrastructure stack:

• Architect cluster management systems handling 100,000+ GPU nodes with heterogeneous hardware
• Develop observability platforms integrating NVIDIA DCGM, Prometheus, and custom ML workload metrics
• Build LLM agents automating hardware procurement, RMA processing, and vendor SLAs
• Implement auto-scaling algorithms balancing research deadlines, cost, and energy efficiency
• Create self-healing infrastructure detecting and resolving node failures in seconds
• Optimize Linux kernel parameters for AI training workloads achieving 20%+ performance gains
• Design firmware update orchestration minimizing downtime across global data centers
• Build network topology discovery and routing optimization for InfiniBand/RoCE fabrics
• Develop capacity forecasting models predicting compute needs 6+ months ahead
• Automate security compliance across bare-metal and cloud-hybrid environments
• Create researcher-facing portals for cluster utilization and job prioritization
• Lead cross-team initiatives improving end-to-end research infrastructure velocity
• Maintain 24/7 on-call supporting production AI services used by millions daily

Qualifications

We're looking for senior engineers with proven track records in large-scale infrastructure:

• 7+ years experience building and operating distributed systems at hyperscale
• Deep expertise in Kubernetes/Slurm cluster management at 10,000+ node scale
• Strong Linux systems programming including kernel tuning and eBPF development
• Hands-on experience with bare-metal provisioning (MAAS, Tinkerbell, or equivalent)
• Production experience managing GPU fleets (NVIDIA A100/H100 clusters preferred)
• Proficiency in Infrastructure as Code (Terraform, Pulumi) and GitOps workflows
• Experience building observability systems (Prometheus, Grafana, custom agents)
• Familiarity with ML platform tools (Kubeflow, Ray, custom schedulers)
• Strong Python/Go/C++ skills for systems-level tooling
• Track record automating complex operational workflows to zero-touch
• Experience with high-performance networking (InfiniBand, 400G Ethernet)
• BS/MS in Computer Science, Electrical Engineering, or equivalent experience

Salary & Benefits

Compensation Range: $220,000 - $380,000 USD base salary + equity + bonus (SF location). Total compensation includes comprehensive benefits package.

• Industry-leading medical, dental, vision coverage (90%+ premiums covered)
• 401(k) with 4%+ employer match, immediate vesting
• Unlimited PTO + 16 weeks paid parental leave
• Hybrid SF work model with catered meals daily
• $10K+ annual learning stipend (conferences, courses, books)
• Full relocation package including temporary housing
• Generous equity grants with 4-year vesting
• Fitness reimbursement, mental health support, commuter benefits

Why Join OpenAI?

OpenAI isn't just building AI products—we're creating the infrastructure that makes artificial general intelligence possible. Your work on the Fleet team directly enables:

• Training runs consuming more compute than most countries' entire grids
• Safety systems preventing catastrophic model failures at scale
• Products serving 100M+ weekly users with 99.99% reliability
• Research pushing fundamental limits of what AI can achieve

Join a team where engineers own entire surface areas—from kernel patches to global orchestration. Work with PhD researchers, hardware architects, and product leaders shaping AI's future. Our culture emphasizes impact, ownership, and rapid learning in humanity's most important technical challenge.

How to Apply

Ready to build the infrastructure powering AGI? Submit your resume and a brief note about your most impactful infrastructure project. We're particularly interested in:

• Experience scaling GPU infrastructure for ML workloads
• Custom tooling you've built for fleet management
• Performance optimizations achieving measurable business impact

Application Process:

1. 30-minute recruiter screen
2. Technical deep-dive with engineering team
3. Systems design interview (cluster management scenario)
4. Take-home project (optional, by mutual agreement)
5. Team fit conversations
6. Offer!

We review applications on a rolling basis. SF-based candidates preferred due to hybrid requirements.