Senior AI Compute Infrastructure Lead - Multi-Provider GPUs

UMA; Paris; Sur place; EUR 120 000 - 190 000; Plein temps

Résumé du poste

UMA is seeking a Compute Infrastructure Lead to own and scale the compute backbone that provisions, schedules, and runs training and data-processing workloads across multi-provider GPUs. You will ensure reliability, efficiency, and researcher velocity as the platform grows from research to production.

This hands-on role combines architecture, ops, and leadership. You will build elastic scheduling, distributed training pipelines, and observability, working closely with researchers and partners

Qualifications

• 8+ years in ML/compute infrastructure, HPC, or large-scale GPU platform engineering.
• Proven track record building and operating infrastructure for large-scale AI model training.
• Deep hands-on experience with GPU clouds and cluster ops.

Responsabilités

• Own compute platform end to end—from provisioning GPU capacity to keeping training, eval, and processing jobs running at high utilization.
• Build a multi-provider management layer to place and fail over workloads across cloud providers and HPC.
• Design and operate the cloud scheduler with quotas, priority, and preemption.
• Stand up a distributed compute framework for training/evaluation on heterogeneous hardware.
• Orchestrate data-processing workloads at scale across CPU and GPUs.
• Deliver virtualized dev sessions (VMs on the cluster) for interactive work.
• Build observability across providers using Prometheus, Grafana, and model metrics.
• Forecast demand, manage costs, and negotiate provider terms while maintaining reliability.
• Help set production-grade practices as we move toward partner POCs and scaling.

Connaissances

GPU clusters

Distributed training

Python

Systems engineering

Observability

Cost optimization

High-performance networking

Vendor coordination

Outils

Slurm

Kubernetes

Ray

NCCL

InfiniBand

Grafana

Prometheus

Description du poste

UMA is seeking a Compute Infrastructure Lead to own and scale the compute backbone that provisions, schedules, and runs training and data-processing workloads across multi-provider GPUs. You will ensure reliability, efficiency, and researcher velocity as the platform grows from research to production.

This hands-on role combines architecture, ops, and leadership. You will build elastic scheduling, distributed training pipelines, and observability, working closely with researchers and partners