DataOps / MLOps Engineer

About the role

We are looking for a DataOps / MLOps Engineer to join a market‑leading 3D scanning company in Bressanone. You will help build and maintain the infrastructure that powers large‑scale computer‑vision and machine‑learning workflows, ensuring data and model reproducibility across the pipeline.

Key responsibilities

• Design and manage storage systems for multi‑TB image data, 3D assets and training datasets.
• Implement efficient data access and movement strategies for distributed training.
• Set up dataset versioning, lineage tracking and experiment tracking (MLflow, Weights & Biases).
• Build and orchestrate ML pipelines for preprocessing, training, validation and model registration using tools such as Kubeflow, Airflow or Prefect.
• Support multi‑GPU distributed training frameworks (PyTorch Distributed, Horovod, Ray) and optimise resource utilisation.
• Maintain artifact storage, Docker registries and model registries, and automate CI/CD workflows.
• Develop tooling to improve developer productivity for ML workflows.

Required profile

• Strong Linux administration skills.
• Experience with large‑scale data storage (object storage, NFS, distributed file systems).
• Familiarity with dataset versioning tools (DVC, Delta Lake or similar).
• Hands‑on experience with ML pipeline orchestration (Airflow, Prefect, Kubeflow).
• Knowledge of experiment tracking platforms (MLflow, Weights & Biases, Neptune).
• Understanding of distributed training frameworks and containerisation (Docker) with CI/CD pipelines.
• Proficiency in managing Python dependencies and environments.

Required skills

• Linux
• Object storage / NFS / Distributed file systems
• DVC, Delta Lake
• Airflow, Prefect, Kubeflow
• MLflow, Weights & Biases, Neptune
• PyTorch Distributed, Horovod, Ray
• Docker
• CI/CD pipelines
• Python dependency management