At RepairPal, I built AI diagnostic recommendations for a marketplace using OpenAI, RAG over repair data, structured outputs, caching, validation, and fallbacks. The end-to-end React, Next.js, and Node.js feature lifted conversion on busy marketplace pages by roughly 8%.
I also improved marketplace performance, cutting Largest Contentful Paint from about 3.8s to 1.7s, and built reliable Redis-backed background jobs that reduced duplicate processing by around 90%. I standardized AWS container deployments and CI/CD to make rollbacks a one-step operation.
Previously at Sanebox and Cadre5, I built React dashboards, Node.js APIs, email automation, accessible government-facing applications, and scalable data workflows. I bring 6+ years of experience shipping and supporting full-stack products, including LLM integrations, reliable APIs, and production systems.
