May 2021 – Present · 4 yrs 10 mos · Seattle, Washington, United States

• AWS Builder Tools
• Identified critical gaps in the Infra as Code safety-analysis service that gates code reviews (CR) across 30K+ CDK libraries
• and all Amazon deployment pipelines.
• Developed and launched a Model Context Protocol (MCP) tool that enriches safety-analysis API findings with workspace
• and change context. This provided real-time, inline feedback that cut support SLA from 3 days to instant for 70k+
• engineers.
• Led development of deadly embrace detection & nested stack analysis at CR time for infrastructure-as-code preventing
• blocked deployments and lowering high-severity rollback/postmortem.
• Amazon Featured Offer(Buy Box)
• Designed and launched real-time customer-intent integration for the ML-driven offer-selection, incorporating
• session-level signals (brand affinity, item condition, delivery speed) to improve relevance and long-term profit at global
• scale (peak ~10M RPS across regions).
• Led a team of 3 engineers and coordinated 9 cross-org engineers to roll out the intent platform across 6 retail programs,
• unlocking ~$100M/year in gross contribution profit and reducing onboarding time for new intent signals.
• Delivered DAG-based parallelization for 50+ downstream service calls—met the zero-latency-regression mandate and
• unblocked delivery; leveraged the result to secure senior-leadership approval for a broader architecture redesign.
• Implemented ZIP-code-level precomputation, clustering 100K+ ZIPs per SKU to cut compute ~10,000×, enable daily
• precompute for 10M+ products, improved accuracy 89%→94%, and unlocked ~$125M/year in incremental ad revenue;
• mentored two engineers and shipped an ultra low latency DAX cache-backed client library to drive adoption.
• Optimized partner team’s ML experimentation platform, unifying treatment configuration into a central service and
• automating promotion from experiment → A/B → release—saving 50 engineer-weeks and 150 calendar weeks.