Senior ML Infrastructure Engineer (Remote, GPU Clusters)

Talantir; Paris; Remote; EUR 90,000 - 150,000; Full time

Job summary

Pathway is seeking a Senior ML Infrastructure / DevOps Engineer who loves Linux, distributed systems, and scaling GPU clusters. You will own the infrastructure for ML training and inference workloads across multiple cloud providers, from bare-bones Linux to container orchestration and CI/CD.

Your home is production infrastructure: clusters, networks, storage, observability, and automation. You will sit close to the R&D team, but your work will directly determine how fast we can train, ship, and

Qualifications

• 5+ years in DevOps/SRE/Infrastructure with production ML workloads.
• Strong Linux systems and networking knowledge.
• Experience with Slurm, Docker, Kubernetes in production.
• Proficiency in Python and common ML libraries.

Responsibilities

• Design, operate, and scale GPU/CPU clusters for ML workloads.
• Build robust ML pipelines for data ingestion, training, and deployment.
• Implement ML-centric CI/CD and automated testing.
• Monitor and troubleshoot training and serving with observability stacks.
• Collaborate with ML engineers to productionize experiments.
• Participate in on-call rotation and incident post-mortems.

Skills

Workload management

Containerization

Orchestration

Slurm

Docker

Kubernetes

CI/CD pipelines

Cloud platforms

Python

ML frameworks

Tools

Terraform

CloudFormation

Grafana

Prometheus

Loki

CloudWatch

SageMaker HyperPod

Vertex AI

Kubeflow

MLflow

Airflow

Metaflow

PyTorch

TensorFlow

Job description

Pathway is seeking a Senior ML Infrastructure / DevOps Engineer who loves Linux, distributed systems, and scaling GPU clusters. You will own the infrastructure for ML training and inference workloads across multiple cloud providers, from bare-bones Linux to container orchestration and CI/CD.

Your home is production infrastructure: clusters, networks, storage, observability, and automation. You will sit close to the R&D team, but your work will directly determine how fast we can train, ship, and