BlackPrint Technologies · Apr 2026
BlackPrint Radar
Multi-Source POI Matching Pipeline
Overview
Engineered a multi-stage entity-resolution pipeline that merged, deduplicated and cadastrally enriched 230,553 commercial and government business records in Querétaro, producing the dataset behind the Radar Comercial platform.
Highlights
- 01
Developed weighted name, address, phone, category and distance scoring that combines Jaccard-containment token matching, adaptive weight redistribution, a known-chain fast path and 384-dim multilingual MiniLM embeddings to match English Dataplor categories with Spanish DENUE labels.
- 02
Matched 6,795 Dataplor–DENUE pairs at a 25 m radius and linked 19,509 POIs to municipal licenses at 350 m using a spatial grid sized at twice the search radius, resolving 215k × 34k candidates in about 3 minutes.
- 03
Enriched POIs through STRtree joins in UTM 14N against 2023 PMDUQ zoning, parcels and condominium units, and shipped a 53-column backend export keyed by an H3 res-12 + BLAKE2b poi_id, covered by an 87-test pytest suite.
Stack
- Python
- Pandas
- Entity Resolution
- Sentence-Transformers
- Multilingual Embeddings
- Shapely STRtree
- H3
- Spatial Grid Indexing
- Pytest
- BigQuery
Working on something similar?
Tell me about the problem and the data behind it. I reply within 24–48 hours.
Discuss a project