• Built a restaurant recommendation agent using RAG architecture and AWS infrastructure from scratch. • Combined RAG with custom prompts for GPT-4o, improving recommendation accuracy from 50% to 90%. • Extracted 5000+ venues via Google Maps API, vectorized with text-embedding-3-small and OpenSearch. • Deployed serverless backend with AWS Lambda and API Gateway, achieving 2-second responses at 100 QPS. • Created an iOS app with SwiftUI featuring real-time chat and one-tap Google Maps integration. • Configured VPC networking with private subnets, NAT Gateway, and IGW to enable Lambda internet access