I turn large-scale location & movement data into decisions — the maps, audiences and models behind real products. Years of building data & location-intelligence systems end to end: data pipelines, geospatial analysis, and the tools that make big data actually usable.
- 🛰️ Building a location-intelligence platform — turning where people go into planning & insight
- 🗺️ Comfortable across the whole path: raw data → pipelines → geospatial analysis → shipped product
- 🧠 Drawn to problems where data meets the real world
Big-file data prep that never runs out of memory — an interactive shell and a one-line CLI. Clean and reshape CSV / Parquet / JSON / Excel files that are too big for pandas. DuckDB does the heavy lifting; kenze auto-sizes memory so your job never crashes. No SQL required — no lock-in either.
pip install kenze # then just run: kenze
Big data, intelligence-ready. A smarter geospatial data format + SDK. K1 is a self-describing, H3-sorted GeoParquet format (
.k1) that replaces "dumb" Parquet; K2 is a DuckDB-powered analysis & sharing layer (.k2). Built for location data that's ready to query the moment you open it.
The open documentation standard for the AI era.
.z1files are AI-native docs — machine-first, around 10× fewer tokens than a README, built on the bet that AI is the new discovery layer. A family of formats (identity, API, changelog…) modeled on how robots.txt, Markdown and schema.org became standards.
