Sateesh Kumar

Designed and maintained multi-layer cloud data pipelines across staging, history, and data mart layers for scalable data integration and transformation workflows. Built Python-based automation to validate and ingest CSV/XLSX files from cloud storage, and implemented automated data quality and reconciliation checks with exception reporting for faster troubleshooting. Developed reusable PySpark functions to support common transformations and optimized PySpark workloads by reducing unnecessary transformations and shuffles. Implemented secure S3 file access using pre-signed URLs and supported production issue resolution.

Built SQL-driven ELT workflows integrating policy, claims, and customer data to support enterprise analytics and reporting. Developed incremental dbt models for premium calculations, claims tracking, and policy lifecycle analytics, with automated post-load validation and monitoring using Python and the Snowflake connector. Implemented RESTful APIs for core HTTP methods and added S3 versioning, lifecycle policies, and error-handling to improve reliability and reduce storage costs.

Deployed a large set of modular dbt models across staging, intermediate, and mart layers to support claims and pharmacy analytics. Implemented CDC-based incremental pipelines to synchronize data changes and created extensive data quality checks using uniqueness and referential integrity rules plus custom SQL assertions. Developed reusable dbt macros for SCD Type 2 processing and hash-diff logic, and used query profiling and clustering analysis to reduce execution time. Added role-based access with row/column-level security and used Python/Pandas for reconciliation and defect investigation with JSON-configurable ingestion and logging frameworks.

Developed ELT workflows to integrate heterogeneous data sources for e-commerce reporting and analytics. Created external stages and optimized ingestion for CSV, JSON, and Parquet datasets, then built extensive SQL transformations using CTEs, joins, window functions, and aggregations. Designed dimensional models for order, inventory, and reporting, and improved performance using clustering and micro-partition pruning. Used Snowflake time travel and zero-copy cloning for development support, troubleshooting, and environment refresh activities.