ML / Computer-Vision Engineer Earth Observation
Location: Stockholm (hybrid) or remote in Sweden
Reports to: Geospatial Lead / Head of Data
Type: Full-time
Role summary
You’ll design, train, and productionize EO-specific computer-vision models (segmentation, detection, temporal) and operational dynamics at industrial sites. Your work goes end-to-end: data, modeling, evaluation, deployment, monitoring, and retraining—integrated into our product platform.
What you’ll do
Build and fine-tune models for multispectral/SAR imagery (e.g., U-Net, Mask R-CNN, temporal networks).
Create reproducible training pipelines (data versioning, augmentation for multi-band/SAR, robust eval suites).
Define domain-aware metrics (m³ volume estimation, per-AOI precision/recall, temporal deltas).
Implement explainability (XAI) and physics-aware validation against ground truth.
Partner with data engineers to ship scalable batch/online inference (tiling/COGs, caching, batching).
Set up model monitoring, drift detection, and scheduled retraining (MLflow/Kubeflow/CI).
Contribute to platform APIs and features for visualization and alerting.
Must-have
3–6 years in ML/CV with hands-on satellite/multispectral and/or SAR data.
Strong in PyTorch or TensorFlow; practical experience with U-Net/Mask R-CNN and temporal methods.
Proven experience building reproducible pipelines and spatial evaluation metrics.
Solid understanding of EO artefacts (clouds, shadows, atmospheric effects) and their impact on models.
Proficient Python; EO tooling familiarity (GDAL, rasterio, xarray).
Nice-to-have
Experience with volume estimation from overhead imagery, sensor fusion, SAR techniques.
Pansharpening, mosaicking, and production deployments in cloud environments.
Cost-aware inference at scale (GPU/spot, tiling/COGs, smart batching).
Why Centersource
Build the core intelligence behind a Sweden-based product that informs real business decisions.
Fast path to production (architecture → pilot → rollout) with clear impact.
Autonomy, focused scope, and quick decision cycles within a senior product team.
How to apply
Send your CV/portfolio/GitHub plus a short note on EO/CV projects you’ve shipped to mai@centersource.io
Subject: ML/CV Engineer (EO).
Working language: English; Swedish is a plus. Applicants must have the right to work in Sweden.
