top of page
APPS
Where technology reaches the people it usually misses
Two products, two very different distances
— one for the elderly neighbors down the street, one for communities on the other side of the world
TEAM SHARESPRING
Shimteoro (쉼터로)
"The path to warmth and coolness"
· 2026 Gyeonggi-do Public Data AI Startup Competition
' WHY I BUILT IT '
It started at a nursing home, fixing wheelchairs. Between repairs, residents kept bringing up the same fear: during a heatwave or cold snap, they had no fast way to know which shelter was actually open, staffed, and safe to walk to. That worry, repeated by person after person, became the question the whole project tries to answer.
' THE PROBLEM '
Gyeonggi-do's senior facility, weather, transit, and pedestrian-safety data all exist — scattered across separate public datasets, in PDFs and spreadsheets no one can use standing on a sidewalk. Existing services each solve one slice: Bokjiro searches welfare facilities but doesn't prioritize immediate weather-safe shelters; Safety Didimdol gives disaster alerts but no senior-specific routing; general map apps optimize for shortest distance, ignoring whether a shelter is even open. "Nearest" isn't the same as "safest," and nothing available connected the two.
' THE APPROACH '
An explainable scoring model weighs six factors — distance, availability, weather fit, transit access, pedestrian safety, and data reliability — into one ranked recommendation, with weights ordered so no single factor like raw distance can dominate the others. Every suggestion is generated with retrieval-augmented generation, so the AI only ever describes facilities that exist in the verified public database — it can't invent a shelter or a phone number.
' DESIGN CHOICES '
Every design decision traced back to who'd be holding the phone. Large text and voice search, because small menus and multi-step search are exactly where older users get stuck. A five-step journey — launch, check conditions, view the top three, get directions, connect for help — kept under four taps end to end. One-tap calling to the shelter, a guardian, 119, or the welfare center, because in a heat emergency, a login screen is a real barrier.
' BUSINESS MODEL & STATUS '
Structured as B2G/B2B2G rather than a paid consumer app — sold to local governments directly, per-municipality. Idea-stage MVP validated against four real scenarios in Yongin and Seongnam before a line of code was written, with a roadmap from a two-city pilot toward all 31 Gyeonggi-do municipalities.
' WHAT I LEARNED '
Handling weather data was my piece of this, and I went in assuming "raw public data plus an API" would be most of the work. It wasn't — the harder problem was deciding how much any one factor should count. A shelter 900 meters away by a safe route is sometimes the right recommendation over one 250 meters away past a pedestrian-accident hotspot, and getting a team of five to agree on how to weigh distance against safety took longer than pulling any of the individual datasets. I came away with a much more concrete sense of what "designing for someone else's context" actually means in practice, not just as a phrase in a pitch deck.
Share Spring (나눔의샘)
A transparent crowdfunding platform, built by students, for society
SELF-FOUNDED NONPROFIT
' WHY I BUILT IT '
The same instinct behind Shimteoro — find a real, specific need and build toward it — didn't stay local. Founding Share Spring as a nonprofit was a way to point that same instinct at communities I'd never meet in person, starting with disaster and climate recovery efforts abroad.
' THE PROBLEM '
Conventional giving rarely shows a donor where their money actually goes or whether it reached anyone. That gap between wanting to help and trusting that the help landed is what keeps a lot of people from giving at all.
' THE APPROACH '
A funding page for each campaign — goal, funded percentage, supporter count, and the people running it, named and visible — paired with a community space where supporters can ask questions and follow a campaign after donating rather than losing track of it. The current campaign, Roots of Resilience, funds a Korean youth volunteer program bringing climate education and landslide recovery support to Aceh, Indonesia, with a $25,000 goal.
' WHAT I LEARNED '
Building the funding side was more straightforward than building trust. A clear goal number and a progress bar are easy to design; making supporters feel like a campaign is still real and still moving, weeks after they've donated, is not — which is why the community forum ended up mattering as much as the funding page itself. It taught me that a platform for giving has to keep proving itself after the transaction, not just before it.
bottom of page