Lead GPU Infrastructure Engineer (Hybrid; AWS, IaC, Scheduling)

Electronic Arts (EA); Redwood City (CA); Hybrid; USD 193,000 - 297,000; Full time

Benefits offered by this job

Paid time off

Sick time

401(k)

Job summary

Electronic Arts is seeking a Lead Infrastructure Engineer to own the GPU fleet and set technical direction for GPU operations. This hybrid role requires three days per week in Redwood City, Montreal, or Vancouver, reporting to the Head of Data and Infrastructure.

You will build the scheduling layer from zero, diagnose GPU and node failures, and drive capacity planning with AWS support. You will also create runbooks and onboarding docs to enable scalable researcher support.

Qualifications

• 8+ years of experience operating production infrastructure with deep AWS depth.
• Experience scheduling, diagnosing, and managing GPUs in AWS.
• Experience operating GPU fleets at 1000+ GPU scale.
• Scripting and automation with Python, PowerShell, bash, or equivalent.
• Infrastructure as code (Terraform or equivalent).
• Familiarity with GPU scheduling/orchestration (Slurm, Kubernetes with Kueue or Volcano, Ray, dStack or SkyPilot).
• Observability practice including Grafana, Prometheus, or equivalent.

Responsibilities

• Own GPU fleet operations across the AWS estate.
• Build the scheduling layer from zero.
• Diagnose GPU and node failures fast and drive hardware replacement through AWS support and capacity-block channels.
• Run researcher support as a first-class product including holding office hours, owning the support channel, and driving recurring causes out of existence with self-service tooling, preflight checks, and documentation.
• Instrument the fleet including utilization, queue depth, job success rate, and cost per experiment metrics.
• Partner with external compute and lab partnerships as a technical contact, and with EA's central infrastructure groups on shared services and escalation.
• Author runbooks, decision records, and onboarding docs.

Skills

AWS GPU management

Python scripting

IaC with Terraform

GPU scheduling/orchestration

Observability (Grafana/Prometheus)

Kubernetes

Tools

Terraform

Grafana

Prometheus

Kubernetes

Slurm

Ray

Job description

Electronic Arts is seeking a Lead Infrastructure Engineer to own the GPU fleet and set technical direction for GPU operations. This hybrid role requires three days per week in Redwood City, Montreal, or Vancouver, reporting to the Head of Data and Infrastructure.

You will build the scheduling layer from zero, diagnose GPU and node failures, and drive capacity planning with AWS support. You will also create runbooks and onboarding docs to enable scalable researcher support.