Data Engineer – Ingestion & Dataset Construction

About the role

Our ML teams train models on datasets derived from large volumes of raw, unstructured data, and model quality depends directly on data quality. As a Data Engineer operating out of Paris, Lausanne, or Zurich, you will own the data layer that feeds the team's models, turning terabytes of raw field logs and public datasets into clean, versioned, training‑ready data.

Key responsibilities

• Ingest, decode, and store raw, unstructured field data (video and sensor streams) into efficient, controlled formats.
• Align multiple data streams temporally and spatially for multimodal training.
• Parse, filter, de‑duplicate, and curate both ingested and public datasets into high‑value data.
• Define data collection and labeling requirements, own dataset‑construction workflow and tooling.
• Build task‑specific datasets, maintain versioning and lineage, and store data in training‑ready formats while managing storage tiering.
• Deliver clean, documented datasets and loading tools to keep model training I/O‑bound.

Required profile

• Degree in a STEM field or equivalent practical experience.
• Proven experience building and maintaining pipelines for large‑scale unstructured data (video, images, sensor data).
• Strong Python programming skills and ability to optimise data loaders for frameworks such as PyTorch.
• Systematic, quality‑minded, pragmatic, and service‑oriented attitude.
• Full commitment to the mission of providing defensive edge technology.

Required skills

• Python
• Data engineering (pipeline design, ingestion, storage, curation, versioning)
• PyTorch (or similar training frameworks)
• Columnar and sharded data formats
• Distributed data processing tools