A modular and extensible Feature Store for Geospatial Data built with Python. It helps data scientists and ML engineers ingest, transform, and serve spatial features for machine learning models.

• 📥 Ingestion from shapefiles, GeoJSON, CSV, and OpenStreetMap
• 💾 Save: Save features as Parquet or Feather using a class-based interface
• 🔍 Duplicate Detection: Prevents saving previously stored geometries
• 📌 Open data integration: Added functionality to fetch Open Street Map data (as base maps)
• 📌 Mapify the data: Visualizes the location data

git clone https://github.com/yourusername/GeoSlate_FeatureStore.git
cd GeoSlate_FeatureStore

pip install -r requirements.txt

python -m examples.simple_demo

• Geospatial Tools: geopandas, shapely
• Hashing & Validation: hashlib, pandas
• Data Storage: parquet, feather, csv

• Ingest and store vector and raster data
• Class-based feature store with duplicate detection
• Open Data base map integration
• Visualize location data
• Feature engineering utilities
• Feature serving via FastAPI
• DVC integration for versioning
• Streamlit dashboard for feature exploration
• PyPI release

Kaustubh Chaudhuri
ML Scientist | Geospatial AI