GPU Fleet Orchestrator for AI Infrastructure

AMD; San Jose (CA); Hybrid; USD 180,000 - 240,000; Full time

Job summary

AMD is seeking an experienced software engineer to build Fleet Manager, a secure control plane for operating large-scale GPU infrastructure. You will design production systems spanning distributed control planes, Kubernetes, GPU scheduling, and developer-facing APIs and tools.

The role sits at the intersection of systems software, cloud infrastructure, and accelerated computing. You will influence how engineers train, serve, debug, and operate workloads on AMD GPUs.

Qualifications

• Strong systems-software development in Rust, C++, or Go.
• Experience designing and operating distributed systems or cloud infrastructure.
• Strong understanding of concurrency, state machines, and failure recovery.
• Experience building reliable services with REST, streaming, WebSocket, or gRPC APIs.
• Familiarity with Kubernetes internals, controllers, or scheduling.
• Knowledge of batch schedulers like Kueue, Slurm, or JobSet.

Responsibilities

• Design and develop Fleet Manager’s distributed control-plane services, APIs, schedulers, and tools.
• Build orchestration for GPU training, inference, and interactive workloads.
• Develop scalable scheduling, quotas, and multi-node workload coordination.
• Implement durable reconciliation, retries, and recovery across PostgreSQL and external systems.
• Integrate Fleet Manager with Kubernetes and related runtimes and storage.
• Evolve Fleet Manager toward multi-environment support (Kubernetes, Slurm, Spur).
• Develop GPU health, diagnostics, and controlled remediation features.
• Build secure multi-tenant infrastructure with strong authentication and least-privilege defaults.
• Improve AI inference reliability, including routing, streaming, and metering.
• Define and maintain stable APIs, data models, and operational procedures.
• Diagnose complex failures across distributed services, GPUs, and networks.
• Develop automated tests and production-readiness checks.
• Collaborate with architecture, driver, security, and ML teams on current/future GPUs.
• Participate in bring-up of new GPU, system, cluster stacks.
• Provide technical leadership through design reviews and mentoring.

Skills

Rust

C++

Go

Distributed systems

APIs

Education

Bachelor’s or Master’s in CS/CE

Tools

Kubernetes

PostgreSQL

REST

gRPC

Job description

AMD is seeking an experienced software engineer to build Fleet Manager, a secure control plane for operating large-scale GPU infrastructure. You will design production systems spanning distributed control planes, Kubernetes, GPU scheduling, and developer-facing APIs and tools.

The role sits at the intersection of systems software, cloud infrastructure, and accelerated computing. You will influence how engineers train, serve, debug, and operate workloads on AMD GPUs.