At Luminor Bank, I introduced Airflow to bring recurring pipeline failures under control and delivered curated banking data as Apache Iceberg tables on Glue Data Catalog and S3. I moved batch workloads to EMR Serverless, cutting compute spend 10–12%, and added automated checks to stop invalid data reaching downstream consumers.
At ACG Group, I made T-SQL reconciliation and reporting 4–5x faster while reducing CPU by 20%, and cut report load time from 20 seconds to one. In my Olist Analytics project, I designed a governed dimensional data product with 107 data tests and about 370 contract checks per CI run.
