Compute Infra Lead: Scale Multi-Cloud GPU Clusters

S27a; Paris; Hybride; EUR 120 000 - 160 000; Plein temps

Résumé du poste

UMA seeks a Compute Infrastructure Lead to own and scale the compute backbone that provisions, schedules, and runs training, evaluation, and data-processing workloads. You will shape a production-scale, multi-provider fleet and ensure reliability and researcher velocity.

You will lead end-to-end compute platform design, manage capacity and cost across cloud providers, and drive observability, tooling, and collaboration with researchers to ship production-ready systems.

Qualifications

• 8+ years in ML/compute infrastructure, HPC, or large-scale GPU platform engineering at a senior level.
• Proven track record building and operating infrastructure for large-scale AI model training.
• Hands-on experience with GPU clouds and cluster operations: provisioning, Linux, high-performance networking, storage for training.
• Built or owned schedulers and distributed frameworks (Slurm, Kubernetes, Ray, SkyPilot) with checkpointing and elasticity.
• Observability, reliability, and cost are core engineering concerns.

Responsabilités

• Own our compute platform end to end—from provisioning GPU capacity to running training, eval, and processing jobs.
• Build a multi-provider management layer to place workloads across GPU clouds and HPC.
• Design and operate the cloud scheduler with quotas, priority, preemption, and dynamic checkpointing.
• Stand up a distributed compute framework for training and evaluation on heterogeneous hardware.
• Orchestrate data-processing workloads at scale to share one reliable compute fabric.
• Deliver virtualized GPU/CPU dev sessions for interactive work on the same hardware used for training.
• Build observability that works across providers (metrics, traces, logs, model metrics).
• Own capacity, cost, and provider relationships as a technical lead.

Connaissances

8+ years exp

Python

GPU clouds

Linux

Observability

HPC

Outils

Slurm

Kubernetes

Ray

SkyPilot

NCCL

Description du poste

UMA seeks a Compute Infrastructure Lead to own and scale the compute backbone that provisions, schedules, and runs training, evaluation, and data-processing workloads. You will shape a production-scale, multi-provider fleet and ensure reliability and researcher velocity.

You will lead end-to-end compute platform design, manage capacity and cost across cloud providers, and drive observability, tooling, and collaboration with researchers to ship production-ready systems.