MLOps Engineer — On-Prem Data Infrastructure

MARSS Group; Nice; On-site; EUR 70,000 - 100,000; Full time

Job summary

MARSS Group is seeking an experienced MLOps Engineer to build and operate the data infrastructure supporting our ML environment in Nice, France. This hands-on role sits at the intersection of data engineering, DevOps and infrastructure, focusing on the data side of the ML platform rather than model training.

You will provision bare-metal and VM environments, automate data workflows, and ensure offline operation with minimal internet access.

Qualifications

• Hands-on Linux server and infrastructure experience.
• Strong scripting with Python for automation.
• Experience building offline or restricted-network environments.
• Proficient with containerized workflows using Docker.
• Familiarity with ML lifecycle tools (MLflow, DVC, ClearML).
• Ability to design and implement scalable ML data infrastructure.

Responsibilities

• Design, configure and maintain servers, VMs and environments for data pipelines and ML development.
• Build and maintain data collection, preparation, validation, versioning and availability infrastructure.
• Automate data workflow from collection through to training triggers without manual steps.
• Configure Linux environments and administer ML and data team servers.
• Deploy and maintain ML/data management platforms like MLflow, DVC or equivalent.
• Operate offline-capable environments with local mirrors and private registries.
• Support dataset/versioning, experiment traceability and monitoring.
• Collaborate with ML, Data and Software Engineers to connect pipelines to training and inference setups.

Skills

Linux

Docker

Python

CI/CD

MLOps

Data pipelines

Tools

MLflow

DVC

ClearML

Job description

MARSS Group is seeking an experienced MLOps Engineer to build and operate the data infrastructure supporting our ML environment in Nice, France. This hands-on role sits at the intersection of data engineering, DevOps and infrastructure, focusing on the data side of the ML platform rather than model training.

You will provision bare-metal and VM environments, automate data workflows, and ensure offline operation with minimal internet access.