Zeurk

Zeurk is the Uber Africa needs. A lightweight, offline-capable ride-hailing solution for African cities enhanced by AI for literacy, computing, and support.

Comment

About the Project

What Inspired Us

Zeurk was born from a desire to build something useful for the real people of Kinshasa. Every day, thousands of people rely on unreliable, inconsistent transportation. I’ve personally witnessed the frustration of waiting under the sun for taxis that never arrive. Combining tech with real-world conditions was the only way forward.

What We Learned

We learned the importance of balancing technical ambition with real-world constraints. We experimented with large AI models, open-source map stacks, and multi-app structures—only to learn that simplicity wins when building for low-resource environments. We also learned how much of an impact a 10MB difference can have on user accessibility.

How We Built It

We used:

  • Expo with React Native for cross-platform support
  • Firebase for the real-time backend and cloud sync
  • SQLite for on-device caching
  • OpenRouter (Mistral) for lightweight AI interactions
  • Cursor AI for development speed and refactoring

All UI was optimized for fast rendering and minimal memory usage. We leveraged offline queuing, local caching, and preloaded content.

Challenges We Faced

  • Storage: Integrating Gemma 2B increased the app size to 4 GB. We pivoted to OpenRouter APIs with distilled logic.
  • Connectivity: We had to ensure the app could fully function in offline or unstable internet conditions.
  • Mapping: Offline open-source maps were considered but dropped due to size and complexity.
  • Backend complexity: Supabase caused too many setup issues and was replaced.
  • User confusion: Separating the driver and rider apps confused users—we unified them.

Every challenge pushed us to innovate for constraint—not against it.

Built With

Share this project:

Updates