Gem

Design Process

Sketches & Wireframes

I began with rough sketches to explore layout options and core interaction patterns before committing to any structure. Early questions I was working through:

• How should users input their preferences without it feeling like a survey?
• What's the right level of detail on a discovery card?
• How do we make filtering feel fast, not tedious?

I iterated across 20+ screens, consistently testing flows with mentors and teammates before moving forward.

Key Design Decisions

Cards over lists.

Early designs used list-based layouts for displaying gems — familiar, but they required reading to differentiate options. I moved to a card-based system with visual metric sliders (affordability, photoworthiness, accessibility) so users could evaluate a location at a glance without parsing text. This directly addressed the 85% of users who described existing apps as cluttered and unmanageable.

Preferences upfront, not buried in settings.

The onboarding flow captures user interests before showing any recommendations, so the very first thing a user sees is already relevant to them. This was a deliberate trade — slightly more friction at entry for a significantly better first impression. Testing validated it: users who completed onboarding engaged more meaningfully with their initial recommendations than those who skipped it.

Cross-checking as the trust mechanism.

One of the most common interview frustrations was not knowing whether a recommendation was genuinely good or just well-marketed. Rather than relying on a single source, GEM's algorithm cross-references multiple signals. I worked with Justin and the developers early to understand what was technically feasible, then designed the UI to communicate this verification to users in a way that felt trustworthy rather than technical.

Social without social anxiety.

The "Friends' Picks" feature lets users see where their social circle has been without introducing likes, follower counts, or public performance pressure. Discovery as a shared conversation, not a competition.

Branding

GEM's visual identity is understated and grounded in order to let destinations take center stage rather than compete with the UI around them.

Usability Testing

Round 1 — Mid-Fidelity

A small group of beta users tested the functional prototype. Feedback focused on overall flow, navigation clarity, and whether the core concept resonated. Mentor designers provided a parallel layer of expert critique.

Round 2 — High-Fidelity

With changes implemented, I ran a second round focused on visual design, interaction detail, and consistency. I kept our personas front of mind throughout, asking at each decision point whether this solved the problems our users actually described.

Key improvements across iterations:

• Simplified onboarding — reduced steps required before a user saw their first recommendations
• Redesigned card layout — moved from text-heavy descriptions to visual metrics (sliders for affordability, photoworthiness, accessibility)
• Saved locations tab — reorganized after users struggled to re-find places they'd bookmarked

The Solution

Feature Walkthrough

GEM simplifies and personalizes adventure discovery. By learning your preferences upfront and applying them consistently, every recommendation feels relevant. The core experience is built around four actions:

Final Prototype

With two rounds of testing complete and iterations applied, the designs came together into a cohesive experience. Here's the result:

What's Next

Expand Testing

Run usability studies with a broader, more geographically diverse group of travelers to validate that GEM's model works globally, not just for students in a single region. Collect longitudinal feedback to understand how GEM fits into real trip planning over time.

Personalization & Machine Learning

Use ML models to improve recommendations over time, learning from saved gems, reviews, and travel patterns. Explore mood-based search inputs (e.g., "adventurous," "cozy," "low-key") that reflect how people actually think when planning.

Accessibility & Global Reach

Add multilingual search and culturally specific recommendations. Build an offline mode for travelers who explore without reliable connectivity.

Key Takeaways

Narrowing scope created focus.

Our original concept was broad, basically a location search tool. With mentor guidance, we honed in on hidden gems specifically. That constraint gave GEM a clear identity and made every design decision easier to evaluate. It also, ultimately, led to winning first place in the fellowship.

User feedback shaped the product more than any assumption.

Nearly every meaningful improvement across our iterations came directly from what users showed us during testing, things we could not have predicted from behind a screen. This process reinforced how much listening matters in the design process.

Cross-functional teams make better products.

This was a genuine team effort, from early whiteboarding to late-night iteration sessions. Collaborating closely with Isabelle, Justin, Gianna, and Nicholas showed me how much stronger a product becomes when every discipline is in the room from the start. As the sole designer, I also learned to articulate design reasoning to non-designers, explaining why a decision mattered, not just presenting it. That habit made both my work and the collaboration sharper.