At Dent•u Global Services, I re-architected a Snowflake semantic layer from transaction-level sourcing to governed aggregate tables, shrinking its knowledge graph from 13.1M to about 600K nodes while preserving query accuracy design.
I built an intent-routed Cortex Agent that gives business users natural-language access to a four-table, 24-relationship data model without needing SQL or schema knowledge. I also corrected a metric definition that caused revenue undercounting, reconciling results to an independently verified $508M benchmark.
In Microsoft Fabric, I designed an automated ETL pipeline using the Medallion framework and reusable PySpark transformations. Its metadata-driven workflow processes more than 25 datasets without code changes, and its error handling and notifications cut failure-response time by 15%.
I also built an IPL data engineering pipeline on Azure Databricks, processing historical data into Bronze, Silver, and Gold layers with PySpark and Spark SQL. My work spans SQL, Python, Snowflake, and data modeling, and I hold DP-700 and SnowPro Core certifications.
