R-EYE-DER helps drivers regain visual awareness through gaze-driven AR.

R-EYE-DER is a gaze-driven AR experience designed to help drivers maintain visual awareness during stressful situations. Built for Raven AR glasses at MIT Reality Hack, the system uses lightweight visual guidance to encourage drivers to scan their surroundings rather than fixating on a single hazard.

When we're scared of hitting something, why can't we look away?

Under stress, drivers can become visually fixated on a perceived hazard, narrowing their attention when situational awareness matters most.

This problem felt familiar to me. When I first got my license, I was terrified of driving on the highway. I would get so anxious that I'd stare straight ahead for too long and sometimes forget to check what was happening around me.

But designing an intervention introduced another challenge: adding more information to a driver's field of view could become another distraction.

How might we redirect a driver's attention without demanding more of it?

The problem isn't always seeing the hazard. Sometimes it's looking at it for too long.
Our research pointed to two attention behaviors that shaped the concept:

Together, they revealed a shared opportunity:
Help users direct their attention to the right place at the right time.

Why gaze-based AR?

Rather than simply warning users about hazards, Raven's gaze tracking allowed us to explore a system that could respond to where users were already looking.
Detect → Evaluate Gaze → Redirect Attention
This created one critical constraint:
The interface couldn't become another distraction.

The interface couldn't become another hazard.
Drivers don't have time to stop and interpret an interface. Every visual element competes with information in the physical environment, so we designed around three constraints.

Information needed to be understood in moments rather than read.

Guidance couldn't require physical interaction while driving.

AR elements needed to communicate direction without overwhelming the driver's surroundings.

What I owned

I led UI/UX and interaction design for the experience, designing the gaze-based interaction flow and lightweight visual guidance system in Figma. I also produced the team's pitch and demo video and helped communicate the longer-term product vision.

🥇 1st Place
Best App for Raven AR Glasses
MIT Reality Hack 2026

R-EYE-DER demonstrated how gaze tracking could be used not just as an input method, but as a way for an interface to understand when a user's attention may need support.

Pocket Tarot is a gesture-driven AR experience for Snap Spectacles that blends constellation exploration with interactive tarot card reveals. The experience places spatial visuals directly into the user's environment, allowing them to discover constellations and trigger a tarot reading through hand gestures.

Working with a collaborator who focused on tarot ritual design and content research, I led the interaction and spatial UX design, translating narrative concepts into an intuitive, hands-free AR experience. The project explores how reflective storytelling can unfold naturally in spatial computing without relying on traditional menus or screen-based UI.

Many AR experiences rely on floating menus and screen-based metaphors that interrupt immersion. For reflective experiences like tarot, this creates friction and makes interactions feel unintuitive.

• Gesture interactions are not clearly discoverable
• Floating UI elements break spatial immersion
• Onboarding for first-time AR users is often unclear

I designed a spatial, gesture-driven interaction flow that guides users from constellation exploration to tarot card reveal. The experience uses pinch gestures and world-anchored prompts to reduce visual clutter and maintain immersion while still providing clear guidance.

(1) Spectacles Camera Detects Pinch Gesture

(2) Pinch Gesture Detected, Interaction Triggered

• First-time AR glasses users
• Users interested in calm, reflective spatial experiences
• Explorers of lightweight narrative AR interactions

• Intuitive hands-free interaction
• Clear onboarding within spatial environments
• Minimal UI that preserves immersion

• Designed pinch-based gesture interaction using Spectacles hand tracking
• Structured experience flow from environment scanning to card reveal
• Reduced reliance on traditional menus in favor of spatial cues

• World-anchored onboarding prompts
• Environment-based constellation discovery
• Card reveal placed directly in user space

• Integrated geo-accurate constellation mapping
• Orientation-aware spatial placement
• Lightweight interaction logic for performance on wearable hardware

• Implemented pinch gesture interaction using Spectacles hand tracking
• Designed spatial UI flow from constellation exploration to tarot reveal
• Integrated geo-accurate constellation map based on user orientation
• Developed onboarding UI and in-world guidance text
• Built working prototype in Lens Studio

My collaborator focused on tarot ritual structure and narrative content, while I translated the experience into spatial interaction design and implemented the AR prototype.

• Conducted user testing sessions on Snap Spectacles
• Received industry feedback from Snap team members
• Identified onboarding and gesture discoverability challenges
• Iterating on clearer spatial prompts and interaction pacing

• Incorporating AI-driven interpretation features
• Improving onboarding visuals
• Refining gesture discoverability
• Enhancing spatial transitions and narrative pacing

Five walls meant five opportunities for something to go wrong.

Samoi Tech's projection installations relied on multiple systems and hardware configurations to distribute content across an immersive environment. Setting up playback and mapping content to individual walls could become difficult to manage, particularly under time pressure.

Which content is playing on which wall?
Operators lacked a clear way to understand wall assignments.

What will happen when I press play?
There was no simple way to preview the configuration before sending content live.

Is every display actually working?
The system needed clearer feedback about synchronization and device health.

So I stopped trying to make one workflow do everything.

Separating these workflows allowed the default experience to remain approachable without removing the controls technical users depended on.

Everything had to have a predictable place.
As the application expanded, I organized its functionality around three primary tasks:

Find and organize media.

Configure what appears across the environment.

Understand whether the hardware is functioning as expected.

The Content Player screen shows exactly what's live at a glance: content source, wall assignment, playback state, and a live preview, all in one place.

Designing within a system that already existed

I worked directly with Samoi Tech's lead engineer across multiple iteration cycles, using internal feedback to understand existing workflows and technical constraints. Each round helped refine the information architecture and interaction model while keeping the designs feasible within the existing projection system.

Main Components

I led UX and interaction design for the application, including information architecture, playback workflows, content organization, system-monitoring interfaces, and iterative design in collaboration with Samoi Tech's lead engineer.

The resulting design consolidated content management, multi-wall playback, and system monitoring into a single interface while preserving different levels of control for different operators.