Data engineer
I build local-first data tools. They let AI agents work with real data, without handing over the secrets inside it.
I am a data engineer. I spend my time on the plumbing most people never see. It moves data from place to place, and keeps it correct on the way.
Lately I build for a new reader of data, the AI agent. An agent needs real tables to be useful. It should not need the names, emails, and card numbers hidden in those tables. Most of my work is on that line.
I like tools that run on your own machine, explain where every number came from, and suggest before they act. The two projects below both start there.
This section is placeholder copy for now. Swap in your own story, background, and links.
“Your data, locally. Your agents, on a leash.”
A local-first data workspace. Query with live SQL, see charts and provenance, and mask personal data before an agent ever reads it. Your data contracts enforce the rules, instead of just describing them.
“Never debug the same data failure twice.”
Diagnose a data-pipeline failure against a version-aware knowledge bank, and get the exact fix with its source. It runs offline, and it suggests rather than acts. It covers Airflow, Airbyte, dbt, Spark, Iceberg, and DuckDB.