Designing for AI That Can Be Wrong
Novah Lee is an AI Product Builder with 10+ years across Samsung AI Center and IBM iX. She designs LLM products that handle uncertainty — confidence architectures, reasoning traces, override flows for high-stakes AI decisions. At Samsung she shipped the Calibrated Trust Zone pattern on a trillion-KRW IT investment platform (NPS 60→72, 30% faster onboarding) and ran the AI Bridge program that converted 40 AI projects to 21 shipped (52% vs industry <5%). She now operates an AI-native Product Lab where 24 builders ship AI MVPs in 9 weeks. Open to US relocation; remote-first roles based in APAC.
AI-native Product Lab — 0-to-MVP System
24 builders, 9-week cycles, AI as production infrastructure — shipped products, not decks.
IT Investment Innovation — LLM Effort Estimation Platform
Designed the "Calibrated Trust Zone" — LLM UX where humans make high-stakes decisions with AI that can be wrong. Trillion-KRW-class IT investment platform. NPS 60→72, 30% faster onboarding.
AI Innovation — 40 Projects to Production (POC→Production Motion)
Designed a 6-step framework that converted 40 AI workshop ideas into 21 shipped production workflows — 52% conversion vs industry <5%. 9.07/10 satisfaction.
Samsung Design System — 100+ Components, 18+ Teams
Built an enterprise design system from zero — 100+ components, adopted by 18+ business systems, 40% UI dev speed up. Samsung Operation Excellence Award (2024).
AI MLOps — RAG Data Pipeline (Docpamin + RAGaaS)
RAG-based document processing pipeline UX — non-ML engineers can now configure AI data flows. Reduced model training prep time, shortened deployment cycles for semiconductor AI teams.
Samsung.com — Unified B2C/B2B/B2E Platform Redesign
Shipped B2C/B2B/B2E unified IA blueprint — integrated navigation, search, content taxonomy across Samsung’s full customer spectrum. Measurable e-commerce conversion lift.
Lotte LCIA — AI Prediction & Cognitive Search System
Cognitive search + recommendation system with Product DNA taxonomy (7–8 categories). Data-to-UX pipeline for new product development workflows.
Lotte Confectionery — AI Prediction System for New Product Development
UX for cognitive search + product recommendation system — natural language and image-based product discovery for new product development teams.
Lotte LOSA — Korea’s First AI Shopping Advisor (IBM Watson)
Led UX for Korea’s first AI commerce chatbot — IBM Watson-powered, 240+ dialog scenarios, 1.5M product entity tagging. World Retail Awards Customer Experience Innovation Best Award. 35% response time reduction.
I design AI products that handle uncertainty. Over 10+ years at Samsung and IBM, I built systems where humans make high-stakes decisions with AI that can be wrong — confidence architectures, reasoning traces, and override flows for LLM-based platforms managing trillion-KRW IT investment decisions. ...