Prototype what you’ll eventually ship
Ideas shouldn’t be judged through static mockups. I build production-quality prototypes in VS Code and Xcode so interactions can be experienced, tested, and refined before engineering begins.
7 years, millions of users. Design, code, and AI models. Currently building the future of voice experience for Microsoft Copilot.
AI products can’t be judged from static mockups. My process blends design, code, and live AI models to prototype, validate, and refine interactions before engineering begins.
Ideas shouldn’t be judged through static mockups. I build production-quality prototypes in VS Code and Xcode so interactions can be experienced, tested, and refined before engineering begins.
Every prototype connects to live models through Azure AI Foundry, Microsoft Graph, and production APIs, testing real conversations, voice, retrieval, and multimodal workflows from day one.
The hardest AI problems happen between prompts and responses. I prototype streaming, memory, interruptions, voice, and multimodal interactions that static screens simply can’t communicate.
I don’t stop at handoff. I contribute production code, raise pull requests, build internal AI tools, and create workflows that help ideas ship faster while preserving design intent.