At ZSCALER, I built a GenAI pipeline that evaluates incident summaries across 10M+ DLP records to classify false positives and insider-risk signals. I also developed an agentic workflow that analyzes Elasticsearch logs, suggests fixes, and creates pull requests, reducing resolution time to 5–25 minutes.
I’ve designed asynchronous batch processing for 1M+ operations and improved evaluation latency from 25 minutes to 2 minutes per record after moving to another model on AWS Bedrock with response caching. Before ZSCALER, I worked on secure data transfers, provisioning automation, observability, and payroll integrations at ZETASUITE and A.D.P.

