Sole designer at Voicebox for 13 months across brand, product UX, design systems, and front-end implementation on an enterprise voice feedback platform.
Due to an NDA, I'm unable to show the majority of design work produced during my time at Voicebox. What follows is an overview of the initiatives I led, the decisions that shaped them, and the work I'm able to share publicly.
Voicebox is a browser-based voice feedback platform for enterprise, letting brands capture, analyze, and act on customer voice across any surface. Deploy it anywhere: website, posters, business cards, digital apps. Customers record audio messages in over 100 languages and send them directly to the brand.
For 13 months I was the sole designer, covering brand identity, marketing, core product UX, and front-end implementation. No design team, no inherited handoff process. I built the visual language, the product experience, and eventually the style layer of the codebase itself. Voicebox is backed by Mark Cuban, presented at NRF in Paris, Singapore, and NYC, and officially partnered with AWS, Box, and Snowflake.
As Voicebox shifted toward enterprise, the existing app wasn't built for it. Over six weeks I audited the IA, identified gaps in how the product handled permissions, sharing, deployment, and billing, and designed a new end-to-end experience covering org management, plans, and payment flows. Delivered as high-fidelity Figma files with all edge cases defined.
The recorder is the core surface of the Voicebox ecosystem, the widget that gets embedded everywhere. I redesigned it from the ground up over a month: new layout, integrated contact form, improved customization controls, motion and animation specs, and responsive and embedding behavior. Also redesigned the in-app recorder customization interface.
Voicebox Actions routes voice messages into third-party services and triggers automations. The legacy UI was generating consistent customer complaints. I identified the failure points through customer feedback, iterated on solutions, and delivered high-fidelity Figma files covering all edge cases. Once engineering built the skeleton, I came in via Cursor and polished the UI directly in the codebase.
The original auth flow predated the enterprise pivot. I redesigned it to support the full setup sequence: account creation, org setup, voicebox configuration, while improving the overall look and feel.
Snowflake required the recorder to support text input alongside audio. I designed the keyboard input flow, extending the recorder's interaction model without disrupting existing behavior.
I started by scoping a custom token-based design system to eventually replace Bootstrap. After a few months, the team made the call to abandon that path. The custom system introduced complexity that wasn't justified for a team moving fast with full Bootstrap familiarity. Instead, I led a codebase cleanup and co-authored a living design.md file with the Head of Engineering: a running reference guiding both developers and AI coding agents on layout patterns, Bootstrap usage, and style conventions.
In my last four months, the workflow shifted. Figma handoffs were introducing ambiguity and slowing things down, so I moved to working directly in the codebase via Cursor, adjusting the design system in code, fixing UI inconsistencies, overriding Bootstrap components, and pushing PRs through review to production. I effectively owned the style layer of the app in code.
I built Voicebox's visual language from scratch: logo, color system, typography, overall design direction. The current logo on voicebox.ai evolved from the first version I designed.
In about a week and a half, I took the product from zero to its first marketing video: developed the concept, wrote the storyboard, found and hired a motion designer on Fiverr, wrote the brief, and managed it through delivery. The video directly helped land enterprise clients and investors.
The founder and I conceived a public directory of business voiceboxes, think Yelp for voice feedback. I built the Webflow front-end and worked with an engineer to dynamically populate it from our database. Result: 261 live voiceboxes from brands including McDonald's, Alibaba, Audi, Boeing, and Canva.
About 20–30% of my time was spent designing presentations for enterprise sales, pitching major retail, hospitality, publishing, and government clients across the US, UAE, and Southeast Asia. Specific names are under NDA.