This repo is a collection of data science + geospatial analytics projects with an emphasis on aerospace, defense (unclassified/OSINT), and environmental intelligence.
I like shipping end-to-end systems: data pipelines → ML models → lightweight APIs.
Focus: geospatial, remote sensing (SAR/EO), trajectory analytics, climate/wildlife
Stack: Python, PyTorch, scikit-learn, FastAPI, GeoPandas, Rasterio, xarray, PostGIS
Tools: Docker, DVC, Prefect, AWS/GCP, QGIS
Reach: open an issue or connect on LinkedIn: https://www.linkedin.com/in/kate-mason99
Rapid change detection for disaster response using pre/post-event SAR.
Data: Sentinel‑1 GRD, FEMA polygons
Methods: terrain correction, log-ratio, Otsu + morphology, interactive map
Metrics: IoU, precision/recall, latency
Repo: (link)
Unsupervised clustering + anomaly detection on aircraft trajectories.
Data: OpenSky Network
Methods: Kalman smoothing, HDBSCAN, feature engineering
Output: interactive map + short report
Repo: (link)
- Packaging reusable geospatial utilities (tiling, masking, CRS ops)
- Spatiotemporal modeling with XGBoost and block cross-validation
- Lightweight MLOps templates (DVC + Prefect + GitHub Actions)
- Portfolio
- Resume: (link)
- GitHub Pages / demos: (link)
Reproducible by default: pyproject.toml or environment.yml, Makefile, fixed seeds, and small sample datasets.
Clear metrics: latency, throughput, cost, accuracy.
Ethics: dual-use awareness and transparent limitations.
