Data engineering
Warehouses and pipelines built to be boring in the best way: tested, documented, observable, and cheap to run.
BigQuery · Snowflake · dbt · Airflow · GCP · AWS · Python
Stockholm, Sweden — Head of Data, Majority
Two decades of shipping software, the last several leading data at a fintech serving diaspora communities. Pipelines that don't page you at night, metrics people trust, visualizations with room to breathe.
Get in touchKatsushika Hokusai, Fine Wind, Clear Morning, c. 1830 — public domain
I've been writing code since 2002 — digital agencies (Framfab/LBi, Knowit, Ottoboni, Creuna), then the startup world as CTO at Toborrow and Head of Product at APPRL. Somewhere along the way, the data became more interesting than the features.
Today I lead the data team at Majority, a US neobank built for migrants and diaspora communities. We turn millions of daily events into models, forecasts, and dashboards that the whole company runs on — from subscriber funnels to regulatory reporting.
My design compass points east: good data work, like good Japanese craft, is mostly about what you leave out. 間 — the deliberate empty space — applies to dashboards too.
Warehouses and pipelines built to be boring in the best way: tested, documented, observable, and cheap to run.
BigQuery · Snowflake · dbt · Airflow · GCP · AWS · Python
Semantic layers and metrics people agree on, then charts that answer the question before anyone has to ask it.
Looker · Semantic models · Forecasting
LLM-powered workflows on top of real data: support agents, document pipelines, and the evaluation work that keeps them honest.
LLMs · MCP · Observability & evals