Senior Designer
ROLE
2 Tech Leads
3 Senior Developers
15 Junior Developers
1 Senior Designer
4 Junior Designers
TEAM
August - November 2025
16 weeks
TIMELINE
TOOLS/SKILLS
Figma, FigJam, SwiftUI, XCode; UX Research, Mobile-First, Developer Collaboration, Dashboard Metrics, Design Leadership
OVERVIEW
How can we design AI systems around experiences users can’t fully remember?
Reverie is an iOS app developed as a project for Georgia Tech’s iOS Club, where student designers and developers collaborate to build original iOS apps and compete against other project teams. By using AI-driven features, Reverie allows users to log dreams effortlessly, generate visual interpretations, and uncover emotional and thematic patterns over time.
Problem Space
Dream journaling is often inconsistent and uninformative. Existing apps typically only rely on users manually logging dreams, offering little insight into patterns, emotions, or recurring themes.
So, how do we design a system that turns fleeting dream narratives into structured, analyzable data?
Solution & Impact
Reverie leverages AI and on-device machine learning to automatically analyze, visualize, and categorize dream content, helping users identify emotional patterns, recurring themes, and subconscious insights from their nightly dream logs.
1st Place in Design
1st Place in Overall App
ROLE & TEAM COORDINATION
As senior designer, I led 4 junior designers and collaborated with 20 developers.
USER RESEARCH
We conducted live, semi-structured user interviews to analyze sleeping and dreaming behaviors.
WHO?
College students and professors in the U.S., aged 18–54, with a range of sleep schedules, dream recall frequency, and journaling habits
WHY?
To identify mental models and pain points around dream tracking and identify unmet needs across different lifestyles and routines
participants said dream memories fade rapidly after waking up, and they struggle to recall details even minutes later
"If I don’t write it down immediately, it’s gone within minutes"
"I’ll wake up knowing the dream was vivid, but by the time I open my notes app, I can only remember one or two details."
participants said visual representations act as memory anchors, helping them to more easily recall emotions, characters, and settings over written descriptions alone
"I don’t remember dreams in sentences , visuals feel closer to how my brain stores them."
"When I see an image or even a color that matches the vibe of a dream, it brings back way more than reading a paragraph I wrote weeks ago."
IDEATION & PLANNING
Turning these key user pain points into targeted opportunity areas:
Low-friction, rapid capture for dream tracking
Recall through emotions, imagery, and fragments
Anchoring memories with visual representations
Analysis through visual or synthesized feedback
User journeys explored how users can capture, reflect on, and explore their dreams.
FEATURE DEFINITIONS & LO-FI DESIGNS
Character Generation
Introduces users to their dream history through personalized character visuals generated from prior dream entries.
Weekly Recap
Aggregates recent dreams into a structured summary that highlights new characters, emotional trends, recurring themes, and weekly statistics.
Frictionless Dream Logging
Log dreams through a fast, interruption-minimizing workflow that supports Siri, text entry, and speech-to-text capabilities.
Retrieval & AI Analysis
Revisit past dreams with AI-assisted interpretation, illustrated storybooks, and the ability to continue incomplete dreams.
AI Insights & Biometric Trends
Visualize dream frequency, length, themes, and emotional patterns using dynamic SwiftUI charts, including mood bubbles and heatmaps. HealthKit data such as sleep stages, heart rate, and activity metrics are overlaid night-by-night to reveal correlations between physiological states and dream activity.
DESIGN TRADEOFFS
List Scannability vs. Visual Context
Users preferred visual cues that support memory recall while keeping the dream archive clean and easy to scan, avoiding cognitive overload.
Automated Insight vs. Personal Sensitivity
We balanced AI-driven interpretation with the subjective nature of personal health data by displaying insights as suggestions rather than conclusions.
Meaningful Animation vs. Visual Noise
Animations were preferred where motion reinforced understanding while keeping other transitions static to preserve clarity and navigation speed.
DESIGN DECISIONS
Immediate dream capture through multiple entry points
Dreams fade within minutes of waking, so we enabled multiple low-friction entry points to let users record dreams wherever and however they wake up.
Visualizing dreams through characters and storybooks
Because dreams are remembered as fragments, emotions, and imagery rather than linear text, we chose to visualize dream content through generated characters and illustrated storybooks. These visuals turn abstract dream entries into tangible, explorable experiences that encourage reflection, emotional and narrative recall, and revisiting past dreams.
Supporting dream reconstruction and retrieval with GenAI
Each dream is processed to generate structured interpretations, emotional signals, and thematic metadata, alongside a Finish My Dream feature that produces a coherent continuation to support recall and improve long-term retrieval.
Designing charts around meaningful patterns
We prioritized visualizing data that users could meaningfully interpret over time, including dream frequency, length, emotional patterns, recurring themes, and correlations with sleep and health metrics. We used SwiftUI charts, mood bubbles, and heatmaps to display trends and relationships at a glance without requiring users to parse raw data.
DEVELOPER HANDOFF
5 Designers + 20 Developers
SHIPPED DESIGNS
Visualize Dreams with Unique Characters
Users explore their dream history through personalized character visuals and illustrated storybooks. CoreML and Stable Diffusion generate dream-inspired imagery, while the Vision framework isolates and structures recurring characters. SwiftUI provides smooth gesture-driven animations, creating an engaging and memory-supportive landing experience.
Dream Logging, Retrieval, & AI Analysis
Users capture dreams immediately after waking via text, speech-to-text, Siri, or Health app–triggered reminders. On-device Foundation Model processing extracts themes, emotional context, and offers Finish Your Dream continuations, while Firestore securely stores structured dream data for retrieval and reflection.
Intuitive AI Insights & Biometric Trends
Users view trends in dream frequency, length, recurring themes, and mood patterns alongside correlated biometric data like sleep stages, heart rate, and activity from HealthKit. Dynamic SwiftUI charts, heatmaps, and mood bubbles make complex relationships intuitive and playful, supporting reflection without cognitive overload.
Interactive 3D Constellation of Connections
Users explore thematic and emotional connections between dreams through a 3D constellation of linked entries. Gesture-driven navigation and SwiftUI animations make exploration fluid, while on-device analysis computes associations in real time, revealing subconscious patterns across the dream archive.
IMPACT & FEEDBACK
We presented Reverie at Georgia Tech's iOS Club Demo Day against 3 other teams and were evaluated by mentors and industry professionals from Apple, Microsoft, and Reddit. Our feedback highlighted the app's innovative visualizations, AI-assisted dream analysis, and intuitive UX.
RUBRIC SCORES
AWARDS
1st Place in Design
1st Place in Overall App
REFLECTION & TAKEAWAYS
Leading Design Across Complex Systems
Guiding the visual, interaction, and AI-driven components across the app reinforced how critical design leadership is in multi-disciplinary teams. I learned that setting a clear vision, aligning engineers and designers, and advocating for the user at every decision point is as important as the designs themselves.
Balancing Automation with User Agency
Working with AI features taught me the nuance of giving users guidance without taking away their control. I learned that thoughtfully framing automated insights as suggestions can empower users while respecting their personal interpretation, a principle I now carry into the rest of my design work.
Designing for Memory-Dependent Experiences
Working with dream recall made me realize how fleeting human memory shapes user behavior and design needs. Reducing friction at capture points and using visual cues not only improved usability, but also deepened engagement, showing me the power of designing for cognition, not just functionality.
Working on Reverie with my team was one of the most meaningful design experiences I’ve had. I loved collaborating with developers and fellow designers, learning from their perspectives, and figuring out how to turn abstract ideas about dreams into something tangible and playful. Seeing users interact with our work and sharing it at Demo Day reminded me why I’m so passionate about designing experiences that feel human, thoughtful, and personal.