I design and scale developer platforms powering Amazon Music’s global experiences. In 2025 I led cross-org initiatives that improved resiliency, cut costs, and accelerated developer velocity across dozens of teams.
I design and scale developer infrastructure powering Amazon Music’s global experiences. My work focuses on distributed systems, developer productivity, and AI-driven automation. In 2025, I led several cross-organization initiatives that improved reliability, reduced costs, and accelerated delivery across dozens of teams.
Key Achievements
•
Lead development of the Oncall Companion Agent, an AI agent that investigates service tickets, diagnoses root causes, and executes routine run-book steps (e.g., rollbacks, reassignments) with human-in-the-loop oversight. Initiated the program after surveying hundreds of engineers and built the de facto benchmarking toolkit for evaluating internal operations agents.
•
Proposed and secured funding for Watchlight, an internal analytics platform measuring engagement and efficiency across Amazon Music developer tooling. Led design and deployment with input from principal and senior engineers. Platform now supports 140+ teams and processes millions of events weekly.
•
Led global reliability program for the Amazon Music Developer API, migrating infrastructure across multiple AWS regions to enable automatic failover and improve reliability. Directed a two-engineer core team and coordinated with Principal TPMs and 45+ service owners, achieving zero downtime during rollout and reducing average request latency by 100 ms for millions of users.
•
Serve as GraphQL schema design Bar Raiser, reviewing 40+ schema change proposals in 2025. Authored org-wide API design standards and mentor engineers on extensible API design.
•
Reduced logging infrastructure costs by $300K annually
•
Actively interview engineers, and work to improve internal guidance on interviewing in the AI era.
Worked on the Data Platform team supporting autonomy, simulation, and machine learning teams with large-scale batch data infrastructure.
•
Co-designed an optimized file format (patent pending) for sensor data to replace ROS bags, reducing storage overhead and significantly improving data processing efficiency across Spark-based workflows.
•
Owned core ETL pipelines for sensor data (petabytes), enabling daily model training using Spark, GCP, and Airflow.
•
Owned ingestion API for multi-modal data across hardware vendors (large variance in format and size), unifying schema definitions and reducing pipeline complexity.
