Led engineering on a 1M+ user product serving roughly 1,000 TPS during peak.

• Built a Python forecasting pipeline that extracted endpoints, pulled traffic history from internal observability systems, and modeled TPS growth, cutting a four-month capacity planning process down to two months.
• Drove a platform migration with percentage rollouts and URL-based state handoff, eliminating full page reloads and improving average page load latency by 40%.
• Re-architected the experience from a legacy server-rendered UI stack to React + Spring Boot and introduced a config-driven rule engine so new device flows could launch through JSON configuration instead of core code changes.
• Built an internal AI knowledge assistant using RAG and a custom MCP server, helping unblock onboarding and mentoring two interns through faster ramp-up.

Worked on identity-aware backend systems and AI-assisted developer workflows.

• Built an identity-aware microservice on AWS Fargate for automated membership validation with secure cross-account isolation.
• Led a cross-region migration for a Redis-backed device-awareness workload in EU infrastructure, writing the migration plan, provisioning the destination Lambda + Redis stack ahead of time, and coordinating phased traffic movement.
• Used VPC peering, async dual-writes, and a one-day TTL overlap window to keep Redis state aligned across both regions, avoid meaningful latency regressions, and eliminate cold-start or cache-consistency issues during cutover.
• Integrated an AI code review assistant into CI/CD with AWS Lambda and a managed cloud LLM platform, reducing manual review cycles by 25% through tiered prompt design.
• Deployed an async job orchestration system for large-scale metadata sync with proactive CloudWatch alerting for authentication anomalies.