VibeCheck
Next-generation dating — advanced AI algorithms for matching partners
- Role
- Product Designer
- Team
- 1 — Product Designer2 — Flutter Developers1 — ML Engineer1 — QA
- Period
- 4 months
- Links
- Open design
VibeCheck was conceived as a modern, next-generation dating app where AI plays the central role. The concept rests on the idea that traditional ways of finding a partner are no longer precise enough for real compatibility. The team wanted a product that analyzes not just the text of a profile, but the user's behavior: which video cards they watch, what they like, how they interact with content. Through deep analysis the app should surface more relevant matches.
Goal
Create an AI-powered dating app — a coherent, comfortable experience on iOS and Android.
I was responsible for
Shaping the core user flows, designing from scratch (0 → 1), scaling the design system (Material), and driving product and UX decisions.
About the project
VibeCheck drops the traditional swipe mechanic in favor of a vertical feed of user video cards, similar to TikTok. The platform's algorithms analyze a broad range of behavioral signals — watch time, reactions, preferences and visual traits — to build a more precise picture of taste and find ideal matches.
Design approach
Since the app was built in Flutter, the key challenge was an interface that works equally well on both platforms. We took Material Design System as the base, but stayed flexible: mandatory native iOS and Android elements were kept where critical, while the rest of the visual system was built on Material. I worked extensively with Material components, adapted them to the product's needs and created custom solutions where the stock elements fell short.
Visual research
Development started from scratch — from the first concepts to finished user flows. The main task was to design a coherent experience that is clear and works the same on iOS and Android. For example, I proposed a “carrot” rather than “stick” for completing surveys — a choice that encourages a more active search for a partner.
Final design
The core user paths were designed: sign-in, onboarding, profile setup, the main feed of potential partners, the recording studio, the profile and its editor, the chat list, chat, settings and system dialogs. Special attention went to the subscription logic: premium access let you message someone you liked directly, without waiting for a mutual like — changing the usual dating dynamic.
Current stage
The project is in active development: most of the user flows are ready, but the product has not yet shipped. In parallel an admin panel is planned, where developers can test the AI, monitor the algorithms' accuracy, manage users and grant premium access.
