• Temporal workflows that orchestrate multi-step AI operations: product attribution, laydown image generation, outfit creation, and outfit copy generation • Backend services in Clojure that integrate with LLM APIs (GPT, Gemini, etc.), manage context and prompts, handle streaming responses, and process structured outputs • Data models and PostgreSQL schemas that support complex AI workflows, vector embeddings, and integration with our existing product catalog • RAG pipelines using Qdrant for semantic search, similarity matching, and context retrieval that powers smarter AI recommendations • ClojureScript UIs for review workflows where humans validate, edit, and improve AI-generated content • APIs and data pipelines that connect everything together, monitoring systems that track AI quality, and the infrastructure that lets us ship fast