Keto Train

The Challenge

Full-stack React Native (iOS/Android) and Node.js app with Gemini exposed through GraphQL for personalized metabolic analysis. Semantic food discovery uses pgvector plus Google embeddings so queries like “leafy greens” surface spinach, kale, and lettuce—pairing instant on-device filtering with async AI similarity ranking. Skia and D3 power the visuals; AWS, Docker, Kubernetes, and PostgreSQL run the backend.

Engineering Highlights

• Gemini LLM integrated via GraphQL for tailored metabolic guidance.
• pgvector + embeddings: hybrid client filters plus server-side similarity ranking.
• Skia and D3 visualizations for high-performance health charts.
• Scalable AWS stack: Docker, Kubernetes, PostgreSQL.