Platform Engineer: GPU Inference & Training

Perplexity; Greater London; On-site; GBP 100,000 - 160,000; Full time

Job summary

Perplexity is building a dedicated platform team to own and simplify GPU-driven training and inference workloads. You will design and run a self-serve compute platform, operate a GPU fleet across cloud providers, and implement scheduling, fault tolerance, and observability to keep workloads healthy and predictable.

You will own Kubernetes-based GPU orchestration, write operators and CRDs, and collaborate with inference and cloud infra engineers to align on architecture and roadmap.

Qualifications

• Deep Kubernetes experience - custom operators, CRDs, and multi-cluster federation.
• You've managed GPU clusters at scale: NVIDIA hardware, CUDA, and networking that makes them fast (InfiniBand or RoCE).
• You've orchestrated compute across multiple clouds (CoreWeave, AWS, GCP, or similar) and understand how different each one really is.
• Strong distributed systems fundamentals: scheduling, resource allocation, and fault tolerance under load.
• You write infrastructure and systems-level code in Go, Rust or C++.
• You've supported both long-running training jobs and high-availability inference services, and you know why they pull infrastructure in opposite directions.
• You own problems end-to-end and do well when the path forward isn't laid out for you.

Responsibilities

• Build a self-serve compute platform. Design and own systems that let teams launch training jobs and run inference services without managing GPU provisioning or provider-specific infra.
• Operate the GPU fleet: provisioning, lifecycle management, reliability, and capacity integration across providers.
• Solve for GPU scarcity: scheduling and placement logic to maximize utilization across providers under real constraints.
• Support long-running training and production inference workloads on the same fleet with high availability.
• Own Kubernetes for GPU orchestration: write operators and CRDs, manage clusters across providers.
• Build fault tolerance, autoscaling, and observability to survive node loss or capacity shifts without manual intervention.
• Set technical direction across teams to shape platform architecture and roadmap.

Skills

Kubernetes - operators & CRDs

GPU cluster management

Multi-cloud orchestration

Distributed systems fundamentals

Go

Rust

C++

Tools

CUDA

InfiniBand

RoCE

Prometheus

Grafana

Weights & Biases

Job description

Perplexity is building a dedicated platform team to own and simplify GPU-driven training and inference workloads. You will design and run a self-serve compute platform, operate a GPU fleet across cloud providers, and implement scheduling, fault tolerance, and observability to keep workloads healthy and predictable.

You will own Kubernetes-based GPU orchestration, write operators and CRDs, and collaborate with inference and cloud infra engineers to align on architecture and roadmap.