Cheryl Platz Portfolio
First designer on Amazon's Echo Look, where she helped integrate Alexa's voice interface into a wardrobe-management device, and later worked on multimodal voice/touch notification design for Windows Automotive. Author of "Design Beyond Devices," a book on multimodal and cross-device experience design.
Location
Remote-first, in the most literal sense — no idea where.
Skills
What makes it work
A breakdown of the choices that make this portfolio stand out.
Framing voice work as a multimodal problem, not a screen-free one
Her Echo Look and Windows Automotive case studies both treat voice as one channel among touch, screen, and physical controls rather than a total screen replacement — a distinction that matters because most real conversational AI products still ship inside apps with visual UI.
Case studies built around "how might we" framing
She opens each project with the open-ended design question her team actually asked, showing the design process before the solution — a structure that lets a hiring manager evaluate her thinking, not just the shipped result.
Teaching output as a public credibility signal
Years of published Medium essays and conference talks on voice UI back up the portfolio's claims; the work isn't just asserted, it's cross-referenced against a decade of public writing the industry already cites.
What Conversational AI Designers can take from this
Specific, actionable tips to apply to your own portfolio — no generic advice.
If your conversational work shipped inside a larger multimodal product, say so explicitly — don't let a case study imply voice was the only interface.
Open case studies with the actual design question your team faced, not the finished flow — it shows judgment, not just execution.
Publish supporting essays or talks alongside case studies; a portfolio backed by a public body of writing reads as more credible than the case study alone.
Case studies
Three of these are voice work at Amazon and Microsoft, each written up with team size, stakeholders and what testing changed. The fourth is the book that turned that work into a method. Cheryl Platz's case studies say plainly when a product was sunset, and what she thinks went wrong.
Echo Look (Product and VUI Design)
A camera-based wardrobe device at Amazon, built as a secret 3-person concept pod that grew to an 80-person team. She was employee number 3 and sole UX designer at first. She drew the storyboards shown to Jeff Bezos for greenlight, and later delivered all of the device's call flows, sample utterances and prompts on the Alexa VUI team.
It covers the full arc of a voice product: early concept, a research study that decided a multi-million-dollar hardware cost, and production VUI delivery. Her closing reflection on why the product failed is candid, not defensive.
15 storyboards covering the key use cases became the basis for Concept Review
Built a delayed Wizard of Oz web server to simulate slow or misrecognized voice responses
Wrote the Alexa Q&A definitions for dress codes like 'business casual'
Alexa: Do Not Disturb
The Do Not Disturb feature for Alexa, covering notifications and Alexa Calling. She mined user transcripts where people had tried to ask for notifications, used them to build sample utterances, and tested audio mocks of the prompts with peers. She calls it her most traditional VUI project at Amazon.
It shows a standard VUI process done well: design from real transcripts, test prompts cheaply, and coordinate sound and light cues across teams.
Sample utterances drawn from real Alexa transcripts, not invented phrasings
Earcons play after the prompt when DND turns on and before the reply when it turns off
Purple LED picked as the lowest-impact color for dark rooms at night
Windows Automotive Notifications
A notification system for in-car Windows, designed for touch, voice and hardware controls. Deep dives with each OS hub's program manager produced a taxonomy of 3 types: Critical, Actionable and App. Designs were tested in a simulator built from Ford Explorer controls, with drivers wearing eye-tracking glasses.
Safety-critical multimodal design, validated against a regulator's standard. The finding that 2 choices was 1 too many shows her letting test data cut features.
Voice prompts used words likely to be recognized in a noisy car, even where they differed from the on-screen text
An 'informational' category was dropped after drivers called non-actionable alerts a distraction
Drivers preferred voice over touch every time, which helped justify later voice work on the platform
Design Beyond Devices: Creating Multimodal, Cross-Device Experiences
A book from Rosenfeld Media on designing for many inputs, outputs and devices, built on her product work from Nintendo DS to Alexa. Her author case study covers the process: first draft September 2019 to April 2020, 10 background interviews, and all the diagrams drawn herself.
It takes years of voice and multimodal project work and turns it into a framework other teams can use. The case study also shows her taking reviewer criticism and rewriting the AI chapter from scratch.
Sample chapters: 'The Spectrum of Multimodality' (Ch. 7) and 'Lost in Transition' (Ch. 9)
AI chapter rewritten after technical review showed it was behind current research
Led to her Multimodal Design Fundamentals workshop series, first delivered at UxLX Masters 2021
More Conversational AI Designer portfolios
Ready to build your portfolio?
Follow the Conversational AI Designer roadmap — skills, projects, and timeline to get hired.