Apr 2023 – Present · 3 yrs 2 mos · Austin, Texas, United States

• Led end-to-end backend design and implementation for a new device workflow in Amazon Japan, enabling Amazon drivers to retrieve lockbox keys via delivery app(NFC); mentored 5 engineers, drove cross-company alignment, and delivered on tight timelines supporting more than 2000+ successful unlocks in production.
• Architected a scalable backfill and classification system for 17M+ address locations using AWS Step Functions, AWS Glue, Amazon Locations API; processed ~20M locations to identify ~375K business locations while optimizing for cost and performance.
• Designed and implemented asynchronous Unlock and Connectivity APIs to support 12-tps using a modular, event-driven architecture with extensible SNS/SQS patterns, testing, and monitoring. The architecture has processed over 3M+ total requests
• Developed and Launched a common BI audit pipeline; processing 500K+ audit events with zero post-launch tickets.
• Owned a no-downtime Redshift cluster migration, by co-ordinating with multiple teams and stakeholders.

Jun 2021 – Apr 2023 · 1 yr 10 mos · Austin, Texas, United States

• Improved App security by partnering with product and security teams to analyze API permission risks, document workflows, and enforce production validations; supported 20K+ requests while blocking disallowed actions.
• Drove a multi-month warehouse firmware initiative by OTA’ing ~1.5K devices pre-ship, resulting in ~$200K in cost savings and reduced downstream operational impact.
• Ramped up an image processing service to 100% production rollout via phased deployment and monitoring; service now processes ~200 images/day and is positioned for reuse by other teams.
• Built and launched a Graph → Redshift audit pipeline for a new graph data model, ensuring no production data loss; pipeline has handled 5K+ records with zero failures since launch.
• Unblocked factory manufacturing workflows by resolving factory-floor issues and fixing bugs in a certificate-signer Lambda, enabling successful handoff of a device manufacturing process.
• Reduced on-call burden by eliminating recurring IoT Rule → DynamoDB/Firehose failures through an async Lambda retry architecture with configurable retries and load testing, resulting in near-zero tickets.
• Diagnosed and fixed a production anomaly-detection alarm impacted by Daylight Savings Time, eliminating recurring false alarms that were paging teams every few days.
• Mentored engineers and interns to successful delivery, including guiding an intern to ship an OTA automation tool and enabling BI engineers to stand up Glue-based pipelines independently.