AI-Native Product UX Design Engineer

5/2026 – offen

w8 is a chat-first metabolic health coach: users log meals in natural language or by photo, and the app forecasts their weight, metabolic mode, and carb budget — explained by an AI coach that speaks their language.

AI-native product architecture. The system strictly separates deterministic computation from generative AI. I built a custom physiological forecast engine (pure, test-validated 5-compartment model: glycogen, coupled water, comfort water, gut, fat) that owns every number and delivers near-exact weight predictions — benchmarked in a custom bake-off harness against LLM-generated and conventional calorie-based models, which it outperformed on forecast accuracy by a wide margin. The LLM layer (OpenAI, strict structured outputs) owns everything language-shaped: intent classification, macro estimation, vision (meal photos, nutrition labels, recipe imports), and a streamed coach voice constrained to use the engine's numbers verbatim. Every LLM call runs against schema-validated contracts (Zod + JSON Schema); a self-growing ingredient knowledge base with read-through caching makes repeat analyses token-free; per-turn cost accounting keeps AI spend controlled.

AI-native engineering process. A core challenge was going beyond "prompting": I set up an enterprise-grade development environment — GitHub with branch/PR workflow, CI gates, changelogs, and a multi-repo ecosystem with tiered knowledge layers (domain knowledge, design system, app) that serve as single sources of truth and are loaded token-efficiently as context into AI-assisted development, raising development throughput by roughly two orders of magnitude. I codified design, development, and workflow standards as reusable rule sets and skills, so AI-generated results stay consistent, high-quality, and aligned with standard engineering processes (hexagonal architecture, testing conventions, commit/release discipline).

Full ownership across the stack: spec-driven development (versioned markdown domain model), own design system with Storybook, Next.js 15 / TypeScript / Supabase (Auth, Postgres with RLS), CI on GitHub Actions.

Softwareentwickler, Git, UX / UI Designer

11/2022 – 5/2023

User Research: Kundenbedürfnisse zur Energie und Kostenübersicht
Design und Konzept für eine Vision zur Präsentation bei den Stakeholdern, um Ressourcen und Mittel freizugeben, um die Ergebnisse dann in die aktuelle App implementieren zu können. Ergebnis: Augenöffner, Wow-Effekt, Prozess Inspiration

User Experience (UX), User Interface (UI)

11/2021 – 2/2023

Portal für die Kundenbeziehung, um Projekte zu erstellen, Lieferungen und Rechnungen zu verwalten, Digital Negotiation, UX Optimierungen, sowie neue Brand Integration. Digitalisierung des Salesprozess. Ergebnis: 32% höhere Portalnutzung, durch verbesserte Usability und nutzerzentrierten Features zur vereinfachung des sales-prozess

User Experience (UX), User Interface (UI)