Data Engineer - Multimodal AI Defense Data Pipelines

Harmattan AI; Vaud; Presencial; CHF 120.000 - 160.000; Jornada completa

Descripción de la vacante

Harmattan AI in Paris, Lausanne, or Zurich seeks a Data Engineer to own the data layer feeding ML models. You will manage terabytes of raw, unstructured data, turning it into clean, documented, versioned datasets so modelers can focus on designing and training models, not data wrangling.

You will join early and influence data processing for deep-learning pipelines, ensuring reproducible results and scalable data delivery to the team.

Formación

• Builde pipelines for unstructured data at scale (video, images, sensor data).
• Strong in Python and data engineering, optimizing data loaders for ML frameworks (e.g., PyTorch).
• Bonus: multimodal sensor data, labeling or dataset construction tooling, and distributed data processing tooling.

Responsabilidades

• Ingestion Pipeline: Ingest, decode, and store raw, unstructured field data from field logs into efficient formats.
• Multimodal Alignment: Align multiple data streams temporally and spatially for training data.
• Curation: Transfer ingested data and public datasets into high-value data, including parsing, filtering, de-duplication and revision.
• Data & Labeling Requirements: Define data to gather and how to label; own dataset-construction workflow and tooling.
• Dataset Construction: Build task-specific datasets for training and evaluation with cross-team collaboration.
• Versioning & Lineage: Version datasets and maintain lineage for reproducible training runs.
• Storage & Formats: Store data in training-ready formats and manage storage tiering.
• Efficient Delivery: Deliver clean datasets and tooling to avoid IO bottlenecks.

Conocimientos

Python

Data engineering

PyTorch

Educación

STEM degree or equivalent practical experience

Herramientas

PyTorch

Dataset versioning tooling

Descripción del empleo

Harmattan AI in Paris, Lausanne, or Zurich seeks a Data Engineer to own the data layer feeding ML models. You will manage terabytes of raw, unstructured data, turning it into clean, documented, versioned datasets so modelers can focus on designing and training models, not data wrangling.

You will join early and influence data processing for deep-learning pipelines, ensuring reproducible results and scalable data delivery to the team.