I build data infrastructure — the kind that lives closer to the user than most people expect.
For the last 8 years at Autobooks I've been the hands-on architect behind a browser-native analytics platform for financial institutions: a full pipeline from legacy bank cores (Jack Henry, FIS, Q2) through Arrow IPC → Parquet → DuckDB-WASM, rendering live financial data entirely client-side with no backend query layer. It's in production, it's fast, and it does things most people assume require a server.
More recently I bolted a Claude-powered AI assistant onto the same stack — tool calling, DuckDB queries, Observable Plot chart generation, internal doc retrieval — all running in the browser via Azure AI Foundry.
- Browser-native analytics — DuckDB-WASM + OPFS as a real production stack, not a demo
- Apache Arrow / Parquet as a data contract layer between backend and frontend
- Schema-driven architecture — one object that defines serialization, joins, rendering, AI context, and access control simultaneously
- Geospatial data — maps, location intelligence, anything that puts data on a surface
- LLM tool calling in production — the gap between "I made a chatbot" and "I shipped an AI feature users rely on"
Interested in data infrastructure, developer tools, and geospatial platforms. Remote, Detroit Metro.