I am a Product Designer @ Five9. Currently building Cortex, a voice AI agent builder platform.
A ground up flagship platform enabling hybrid agentic and deterministic systems for the world’s largest enterprises. This project is actively being developed at the moment.
An app that uses the Claude API to generate archetype-based critiques of UI screenshots. Built for my team at Five9 to compress the gap between design iteration and user research — instead of waiting weeks for a usability session to surface obvious failure modes, designers can pressure-test screens against grounded user mental models in minutes. Each archetype is built from real interview data and prompted toward distinct decision-making patterns, so the output isn't generic "user feedback" but specific predictions about what this user would notice, ignore, and misinterpret. Presented to leadership including the CPO; now used by designers on my team mid-Figma, before review.
Internal tool — UI omitted. Happy to walk through live.
A working prototype for a problem I kept hitting personally: there's no good way to annotate audiobooks the way you'd annotate a physical book. Highlights and marginalia break down in audio, you're left with timestamps that lose meaning a week later. The core move in this prototype is voice: you mark a moment by speaking, and your voice itself becomes the bookmark. It is a captured fragment that holds the thought, the context, and your reaction in the same modality as what you were listening to. No pulling out of the experience to type. The design problem I cared most about was keeping the listener in flow: annotation should feel like a small extension of listening. Self-initiated; built earlier this year.
