← All work
Kioo
AI beauty technology with skin-aware virtual try-on
Kioo is my own product: an AI-powered beauty technology platform that does what generic try-on apps cannot. It respects African skin tones, undertones and textures, and recommends products that actually suit the person in front of the camera.
- 74,200
- lines of Dart
- 426
- tests
- 17
- n8n workflows
Why
Virtual try-on, recommendation engines and beauty AI are largely trained and tuned for skin tones that are not African. The result feels uncanny at best and alienating at worst. Kioo is built from the skin tone up.
What it does
- Flutter for cross-platform mobile.
- Firebase for auth, real-time data and analytics.
- Google ML Kit face-mesh for on-device facial landmark tracking.
- Gemini for product recommendation reasoning; Pinecone for catalogue similarity search.
- 17 n8n workflows for catalogue ingestion and vendor operations.
What I led
- Product architecture and the try-on rendering pipeline.
- Skin-tone calibration and the personalisation model.
- Catalogue ingestion and the vendor partner workflow.
- The brand and how Kioo speaks to its users.