Data Scientist

The client leverages satellite-derived observations and other auxiliary data to develop geospatial flood risk models, requiring strong hydrological expertise to extract reliable signals. In this project you will lead the effort to extract reliable hydrological signals from diverse, often noisy, geospatial datasets to support the client's expansion across the United States.

Start: whenever possible, flexible
Duration: end of year, extension possible
Location: Espoo or Valencia
Work type: Hybrid
Language: English
Allocation: 100%

Role description:
This is an outcome-oriented role. You are responsible for the quality of the features and the integrity of the labels that power the client's training pipelines. We expect a balance of an experimental mindset for feature discovery and a 'built-to-last' discipline for model validation. By utilizing AI-augmented workflows, you will maintain high-velocity delivery while ensuring our risk scores remain statistically meaningful and grounded in physical reality.

Your day-to day responsibilities:

• Feature Development: Translate environmental, terrain, and hydrological data into model-ready features grounded in physical process understanding; work closely with other hydrological subject matter experts to validate that signals reflect real-world behaviour
• Observation Quality: Assess, extend, and improve the quality of satellite-derived observation labels, investigating noise, bias, and coverage gaps, and designing labelling approaches that reflect confidence in the data
• Experimentation: Design and run experiments to understand what drives predictive performance; distinguish genuine signal from statistical artefact
• Physical Interpretability: Ensure model outputs are explicable in domain terms, the kind of scrutiny that holds up in front of scientists and domain experts, not just benchmarks
• Collaboration: Work closely with Data Engineers to define robust, scalable data and labeling workflows, and with ML Engineers to ensure features, labels, validation criteria, and model evaluation are scientifically rigorous and production-ready

Requirements (must haves):

• Education: Master's degree or higher in hydrology, geomorphology, geoscience, environmental science, civil engineering (water resources), or related quantitative field
• Experience: 5+ years of professional industry experience in data science with geospatial or environmental data, flood modeling, hydrology, climate risk, natural hazard assessment, or remote sensing analytics
• Domain Knowledge: Deep understanding of flood, extreme weather, and hydrological processes; able to reason about what the data represents physically, not just statistically. Familiarity with terrain analysis, catchment hydrology, or drainage network characterisation
• Geospatial Data: Hands-on experience working with raster and vector geospatial datasets, transforming raw environmental data into analytical features
• ML Proficiency: Solid experience with supervised learning for both geospatial and tabular data, training models, interpreting results, and running controlled experiments
• Python: Python-fluent; writes clean, testable analysis code, not just one-off notebooks
• Data Quality: Experience working with observational data that is noisy, spatially biased, or incomplete, and knowing how that affects model behaviour
• Modern Tooling: Pragmatic use of AI tooling (Cursor, Claude, Copilot) for data exploration and analysis acceleration

Nice to haves:

• Experience contributing to a shipped product, software engineering practices, version control discipline, code review, CI/CD, not just notebooks and papers
• Familiarity with US government geospatial datasets (FEMA, USGS, NOAA) PostGIS, AWS, or Databricks experience
• Remote sensing literacy, SAR, or optical environmental mapping
• Climate risk, insurance, or catastrophe modeling vocabulary

Please note that the client will do a security screening (incl. SUPO, where required) for the chosen candidate.

Interested? Please contact Lisa_Witted in Slack / lisa.sandstrom@witted.com