Visa

At Visa — the network behind hundreds of billions of dollars in annual payment volume — I work on the full-stack data pipeline and quantitative infrastructure team, architecting AIMS (Active Incident Management System) from a greenfield codebase. AIMS replaces slow, manual L2 escalation with an autonomous orchestration layer that detects, classifies, and routes production incidents in real time across Visa's payment infrastructure.

The core of the system is a proprietary routing engine paired with ML-based triage and real-time classification models that operate directly on Visa's production data lake payment telemetry. Instead of a human paging through dashboards, AIMS ingests high-volume event streams, correlates anomalies, infers likely root causes, and dispatches closed-loop remediation playbooks — dramatically compressing the window between detection and resolution.

I design high-throughput ingestion and enrichment pipelines over massive financial data lake volumes, including autonomous knowledge-graph construction that links services, dependencies, and historical incidents into queryable root-cause inference paths. On the platform side I build the web services and APIs that expose this intelligence to on-call engineers and downstream automation, and implement production-grade stream processing for real-time incident enrichment, anomaly correlation, and automated remediation workflows.