The Mission
At Kestrel, we are building the future of scalable autonomy across domains. Our systems enable commercial and national security customers to sense and act in support of complex missions without significant autonomous system expertise. This technology bridges domains (e.g., space, air, land, sea) and fields (e.g., AI, orbital dynamics, geospatial), requiring a team that has both breadth and depth in these related topics. The Role
We are looking for an individual contributor to architect and deploy production-grade subsystems that power our cutting-edge products for both commercial clients and critical national security missions. As a Kestrel Data Scientist, you will lead the development of high-fidelity data processing pipelines, traditional AI/ML models, and wrapping them into tools for next-generation Agentic AI. You will work at the bleeding edge of geospatial science. This role requires adaptability, engineering rigor, and a deep respect for the mission. Core Responsibilities:
- Architecture: Lead the end-to-end creation of production-ready AI/ML and agentic subsystems.
- Geospatial Engineering: Build robust preprocessing pipelines for raster and vector data (using GDAL, SNAP, rasterio, PostGIS).
- Innovation: Integrate emerging technologies, from Geospatial Foundation Models (GeoFMs) to autonomous route planning, into a cohesive operational picture. You Must Be: A master of foundational geospatial: Proven experience building geospatial data preprocessing pipelines (raster and vector) using the modern geospatial stack (GDAL, SNAP, rasterio, PostGIS, etc.). A specialist: You must have production-grade experience in at least two (2) of the following domains:
- Traditional GeoAI/ML: Object detection/land classification (scikit, OpenCV, etc.).
- Geospatial Foundation Models: Utilizing GeoFMs like AlphaEarth, Prithvi, or Clay.
- SAR & Interferometry: Pipeline development (e.g., isce2).
- Hyperspectral Imaging: Processing (e.g., SPy).
- RF Propagation: Modeling signal data (e.g., CloudRF).
- Digital Elevation Models: Handling DEM in multiple formats (e.g., DTED).
- Autonomous Systems: Route planning for unmanned systems (e.g., Auterion, ArduPilot).
- Optimization: Complex scheduling for satellite constellations or aircraft.
- Economics: Models for resource allocation such as auction, bidding, pricing, waiting. Must be a US Citizen Bonus:
- Based in the Colorado Springs, CO or Washington, DC (working in customer sites in Crystal City, VA and Herndon, VA) areas
- Active Top Secret // Specially Compartmented Information (TS//SCI) clearance with counterintelligence-scope polygraph (CSP) or full-scope polygraph (FSP)