Social Discovery & Local Exploration Platform
Cross-platform mobile application enabling social discovery, personalized collections, and effortless sharing for local exploration and community-driven recommendations.
The Challenge
Pingea's founders had a clear product vision: a social layer on top of local exploration, where people could save places they loved, organize them into shareable collections, and follow curators with good taste. The challenge was translating that into a mobile experience that felt social without being noisy, and local without being generic. Technically, the recommendation engine needed to balance proximity, personal preference history, and friend activity signals in real time across a map interface that felt fluid even with hundreds of pins loaded.
Our Solution
We built iOS and Android apps in React Native sharing a GraphQL backend. The map view uses a clustering algorithm to keep pin density manageable at every zoom level, with smooth transitions between cluster and individual pin states. User collections are stored with rich metadata — tags, notes, cover photos — and can be published publicly or shared with specific followers. The recommendation feed combines a user's saved place categories with friend activity and location context to surface relevant suggestions. Push notifications are triggered by friend saves within a user's configured geographic radius, creating organic discovery moments.
Our Approach
Graph-Based Social Layer
Designed the data model around a social graph where users follow curators and their collections. Built GraphQL resolvers that efficiently hydrate feeds with friend activity, handling follow/follower relationships at scale using PostgreSQL with indexed graph traversal queries.
Map & Clustering Engine
Implemented a geospatial clustering algorithm on the client that groups nearby pins into clusters at lower zoom levels and expands them as the user zooms in. Used Mapbox GL for smooth rendering performance. The backend pre-computes pin clusters at fixed zoom levels during off-peak hours to reduce runtime computation.
Personalized Recommendation Feed
Built a recommendation engine that blends three signals: the user's own saved place categories (preference inference), friend activity (saves and new collections), and the user's current location. Results are ranked and refreshed on a short TTL, cached in Redis for fast feed delivery.
Collections & Sharing
Built the collections feature with privacy controls (public, followers-only, private), deep link sharing that renders a web preview when shared externally, and a native share sheet integration. Each collection has a canonical web URL that works even for users without the app installed.
"The map experience is exactly what we had in our heads from day one. The team figured out how to make it feel effortless in a way previous contractors never managed to."
Let's Build Together
Have a similar challenge? We'd love to hear about your project.