Geospatial Data Scientist – Yield Forecasting & Trade Signals
Degas
- Amsterdam, Noord-Holland
- Vast
- Deeltijds
This is a remote-first, contract-based role, open to top-tier talent anywhere in the world, with optional travel to Degas' HQ in Tokyo, investor locations, and regional field operations in Ghana and Côte d'Ivoire.About DegasDegas' mission is “Changing people's lives, dramatically.” We are an innovative AI technology company focused on developing scalable, AI-driven global products that address the world's greatest challenges. Alongside our most advanced tech products, we also operate deeply rooted field initiatives that increase the income of smallholder farmers in Africa and beyond — combining expertise in data aggregation, AI credit scoring, satellite observations, financing, and carbon markets.As part of our next strategic evolution, Degas is launching a new global business division that integrates proprietary agricultural yield intelligence with capital markets — starting with soft commodity trading. This marks a bold new chapter in how we scale impact and create financial value from our AI-driven insights.About the roleDegas FM is seeking a Geospatial Data Scientist to refine cocoa-specific agro-yield prediction models and transform them into structured, backtested trade signals. The successful candidate will work with NDVI, rainfall, and disease stress layers, using Degas FM's satellite-derived data to calibrate anomaly detection and build signal triggers linked to cocoa futures market behavior.This role is ideal for a specialist at the intersection of remote sensing, ag forecasting, and trading signal engineering. Your work will directly enable the firm's internal arbitrage strategy by turning environmental data into quantitative insights actionable in ICE markets.Key Responsibilities
- Signal Development & Engineering: Convert NDVI trends, rainfall deviation, and crop stress indicators into structured yield anomaly signals. Develop automated signal pipelines including: preprocessing of satellite and geospatial data, regional anomaly scoring, threshold-based triggers for trading entry/exit, and signal output formatting for the trading desk.
- Model Calibration & Backtesting: Align signals with historical cocoa price movements using time-series correlation, lag analysis, and seasonality. Maintain and refine a lightweight backtesting module to evaluate model accuracy, false positives, and market sensitivity. Define signal precision metrics (e.g., hit rate, average return per signal) and improve them iteratively.
- Cross-Functional Collaboration: Work closely with the cocoa trader, Degas FM data team, and field intelligence teams to align model assumptions with on-the-ground realities. Integrate regional knowledge (e.g., harvest windows, rainfall anomalies, or disease outbreaks) into signal refinement cycles.
- Signal Monitoring & Maintenance: Ensure consistent pipeline performance, data freshness, and signal integrity over time. Implement automated alerts for model drift, data anomalies, or invalid triggers. Maintain a minimum viable monitoring dashboard or reporting interface.
- Knowledge Sharing & Documentation: Document signal methodologies, assumptions, and known limitations in a clear, accessible format. Share insights regularly with non-technical stakeholders to support broader organizational understanding of yield-to-price dynamics.
- 3+ years of experience in geospatial or satellite-based agro-intelligence modeling (e.g., NDVI, precipitation, disease spread).
- Strong Python skills: pandas, NumPy, xarray, and time-series analysis libraries.
- Proven ability to build forecasting models and signal classification tools for crop yield anomalies.
- Familiarity with cocoa or other tropical perennial crops is a strong plus.
- Prior collaboration with quant traders or basic understanding of commodity markets is highly desirable.
- Accuracy- and outcome-oriented; thrives on impact more than documentation.
- Able to work independently and iterate quickly with minimal supervision.
- Comfortable with ambiguity, lagged data, and low-process environments.
- Cross-functional communicator: can translate statistical insights into strategic recommendations.
- Values transparency, scientific rigor, and collaboration.