At Liven (Nomni AI), I built a parameterized Databricks pipeline that processes 1 TB of data daily across several regions, from raw S3 files through PySpark transformations into silver tables.
I also created a PySpark observability workflow that compares MongoDB source counts with Unity Catalog tables and alerts the team to mismatches. For Liven and Zeemart reporting groups, I replaced manually run jobs with a reusable, metadata-driven refresh process, reducing new-group onboarding from 7 business days to 2.
At Cognizant Technology & Solutions, I built a Python/PySpark data consistency framework that reduced source-to-target discrepancies from 46M rows to 0.18M and cut batch runtime from 18 hours to 7. I also developed reusable ETL frameworks and orchestrated ingestion and transformation jobs with Oozie.
