Engineer first, data leader by conviction.

I spent my first decade as a professional engineer in Canadian energy — production operations, development engineering, and the regulated reserves work where a spreadsheet error is a board-level problem. That world taught me what production-grade actually means: numbers people bet careers on, audited yearly, no hand-waving.

Berkeley's MIDS program (completed nights and weekends, mid-career) turned a long-running data habit into a discipline. Since then I've built three generations of data platforms — Tableau-as-warehouse, then dbt + Snowflake, now Dagster + Polars + DuckDB — each one still running at the company that paid for it.

The last three years changed how I work more than the previous ten. I build AI-native now: LLM-in-the-loop pipelines, structured-output systems, MCP servers, and Claude Code toolchains that multiply what one engineer ships. twochannel.ai is the proof: a 18k-product catalog one person maintains because the pipelines do the work of a team.

Today I'm Product Manager for Stack Maps and the Public Data team at StackDX, coordinating five engineers across two countries while still shipping in the codebases the roadmap depends on.

The pattern across all of it is the same shape of problem: a company sitting on real data with nothing underneath it, or a platform that has stopped being extendable. At Canlin that meant thousands of wells spread across systems that had no idea the others existed, and no data hire before me. StackDX was a legacy stack that had to come out while it kept serving customers. With twochannel I did it to myself. All three took years rather than months, and in each one the order I did things in mattered more than any particular technology.

Away from the keyboard: Calgary, the Rockies, and a long-standing meditation practice.