Toni Kumari

Data Engineer

At Zelestra, I engineered an automated, parameter-driven ingestion pipeline across multiple solar plant sites (150MW+ capacity), extracting ~200 GB/month of high-frequency SCADA telemetry via AWS DMS using CDC and full-load replication. I personally built fault-tolerant extraction workflows with checkpointing, automatic retry-and-backoff, and automated post-upload cleanup. In the transformation layer, I developed PySpark workflows to aggregate raw sensor tags into standardized 15-minute datasets—applying custom aggregation and meter rollover corrections that reduced data mismatches by 93%.