At Novacard, I own a 3 TB consumer-credit warehouse and about 50 automated ETL/ELT pipelines serving risk, CRM, finance, and marketing. I rebuilt a full recompute over 162M CDC rows as an incremental design, targeting runs of 10–30 minutes instead of 90–120 minutes.
At Luchi (formerly BestDoctor), I designed an S3 lakehouse with Trino, Apache Iceberg, and dbt for regulated medical and insurance data. I built the governed dbt layer and led the migration from Airflow to Dagster, with key reports running 40%+ faster.
At Reckitt, I rebuilt a regional data platform on PostgreSQL, Airflow, and S3 after Azure exited the market, reducing infrastructure cost by 70% without business interruption. Earlier, at Epson Europe B.V., I automated reporting and ETL, cutting manual data handling by 90%.

