Translations
Real-time multilingual conversations on smart glasses
Defined and owned the conversation design strategy for the translation domain on Ray-Ban Meta and Ray-Ban Meta Display. Scoped and led the interaction model across three core features: Live Translation, On-demand Translation and Visual Translation, establishing the voice-first framework that scaled across modalities and devices.
Hands-free experience
Translation is a high cognitive load task. Users are processing language and meaning in real time while staying present in a conversation. Any friction breaks the flow. The voice interaction had to feel invisible, fast enough to keep pace with natural conversation.
We designed three distinct translation modes to match different real-world scenarios: live conversation, on-demand queries, and visual text, each optimized for speed and minimal cognitive load.
Live Translation
Designed the end to end voice interaction flow for live translation, defining how users initiate, manage and end translation sessions entirely hands-free. Partnered with another designer for the Meta app experience when users needed to use their phones to set up or manage translation sessions.
Hands-free puts human connection at the forefront rather than tech. Using glasses to facilitate a two-way conversation rather than a phone or translation device lets users maintain eye contact and build a more natural connection.
The team used multiple methods to test how live translation would work in real conversations. As a native Spanish speaker, I played a key role in testing the Spanish models and providing feedback. Initially the plan was to ship a single Spanish model, but through extensive testing I surfaced that it didn't work across regions. This led to the team building separate models for Latin American Spanish and Spain Spanish, with plans to expand to more regional variants.
Live Translation was demoed on stage by Mark Zuckerberg at Meta Connect 2024. I was directly involved in preparing the demo: on-stage testing, providing multiple versions of what to demo based on my Spanish expertise, and preparing Brandon Ramos, the guest speaker who demoed the feature live with Zuckerberg. As a native Spanish speaker, I was the right person to help him understand how the feature worked and feel comfortable on stage.
Shipped Live Translation on Ray-Ban Meta, enabling real-time multilingual conversations hands-free. The feature was demoed live by Mark Zuckerberg at Meta Connect 2024 and became one of the most visible launches for the glasses platform. My Spanish expertise directly shaped the demo content and on-stage preparation, contributing to a successful global debut.
Visual Translation
Visual Translation lets users take a photo with their glasses and ask Meta AI "What does this say?" The assistant reads and translates the text in the image, entirely hands-free.
A key scenario was traveling and looking at a restaurant menu in another language. You need to understand what you're ordering, maybe you have allergies, and you want answers fast. But translating an entire menu via voice creates too much audio. The cognitive load becomes overwhelming when the assistant reads every item aloud.
We had to be very deliberate about what goes through voice versus other modalities. On Ray-Ban Meta Display, the challenge was even sharper: reading every piece of information on a small display felt like too much. We designed the system to be smart about what to speak aloud versus what to surface on the display, keeping the experience useful without overwhelming the user.
Speak the most relevant translation aloud, like what a specific dish is, rather than reading the entire menu.
Leverage the display for additional context that would be too much to speak, letting users scan at their own pace.
Visual Translation was shipped and streamed at the Super Bowl in 2025 on Ray-Ban Meta.
On-demand Translation
On-demand Translation lets users ask Meta AI how to say something in another language, like "Hey Meta, how do you say where is the nearest train station in Spanish?" The assistant responds with the translation spoken aloud, giving users a quick, hands-free way to learn phrases in real time without interrupting what they are doing.