Sumanth Gaddam

• Designed and operated batch data pipelines to convert transactional data into analytics-ready datasets.
• Developed PySpark workflows with schema enforcement and business-rule validations for data integrity.
• Implemented incremental processing and historical backfills, enabling safe reruns without duplication.
• Constructed fact and dimension tables for accurate trend and point-in-time analysis.
• Delivered Gold-layer datasets that powered executive and operational Power BI dashboards.
• Reduced pipeline reprocessing failures by approximately 25-30% through idempotent design principles.
• Processed transactional datasets up to 20 GB per run with production-grade reliability.
• Applied distributed data engineering best practices across various cloud platforms.

Problem Statement
Lack of a reliable, scalable data platform to support high-volume banking transaction analytics and historical point-in-time reporting, resulting in reconciliation issues and reduced trust in KPIs.

Solution Provided

• Designed a Medallion (Bronze–Silver–Gold) Lakehouse architecture using Microsoft Fabric.
• Built PySpark ingestion and transformation pipelines with embedded data quality and schema validation.
• Implemented SCD Type 2 customer dimension and analytics-optimized star schema.
• Developed idempotent Gold-layer pipelines using partition-level delete-and-insert strategies.
• Added audit, reconciliation, and validation controls across pipeline stages.

Business Outcome

• Improved data accuracy, consistency, and trust in executive and regulatory reporting.
• Reduced reconciliation issues and reporting discrepancies.
• Enabled faster, confident decision-making through reliable KPI dashboards.
• Established a scalable foundation for future analytics and compliance use cases.