Context
The problem
Forest-rights claims mostly exist as scanned paper. They're slow to check and almost impossible to compare between regions.
Approach
How I built it
OCR pipelines read and clean the scanned records, so nobody has to type them in by hand.
Simple statistics and ML checks flag records that look unusual, which speeds up checking claims.
Map dashboards backed by PostgreSQL show claim status and land use for each region.
Result
What came out of it
100+ paper records turned into data
Odd-looking claims flagged early
Live dashboards by region