Machine Learning Engineer, Earth Observation

About the team

Terravek's data science group owns the models behind change detection, defect classification and crop stress. We publish model cards with every release, including the cases where the model does not work.

What you will do

• Take research prototypes from Terravek Labs to production models
• Own training pipelines, evaluation harnesses and the published model cards
• Work directly with product managers on where a model should and should not be trusted
• Contribute to the uncertainty quantification work with Terravek Labs

What we are looking for

• Strong applied ML background with production experience, not only research
• Understanding of remote sensing physics — or genuine willingness to learn it properly
• Comfortable with the idea that a confident wrong answer is worse than an uncertain right one
• MSc or PhD in a relevant field, or equivalent practical experience

Skills and technologies

These are the things we cannot train quickly. Everything else we can.

• PyTorch
• Python
• Remote sensing fundamentals
• MLOps
• Distributed training
• Geospatial data formats

Useful, not required

• Publications in remote sensing or computer vision
• Self-supervised learning
• Hyperspectral data
• Ray or Kubeflow

What we offer

• 8% holiday allowance
• 27 days annual leave plus public holidays
• Pension via Stichting Pensioenfonds Techniek
• Hybrid working, 2 office days a week
• €1,500 annual learning budget
• NS Business Card or bicycle scheme
• Employee share participation plan after 12 months

Full benefits by country