Nura × Real-Time AI Skin Analysis

Bringing AI-powered skin analysisto the retail floor

Product DesignMachine Learning SystemsFull Stack Engineering
Nura totem interface

Role

Founder, Designer, Engineer

Timeline

September 2025 – Present

Team

Solo project

The Problem

In retail stores, skincare recommendations are often guesswork. Consultants rely on subjective visual assessment, customers struggle to describe their skin concerns, and product decisions become inconsistent. The result: low trust, low conversion, and no data continuity between users and retail teams.

The Opportunity

Build an AI-powered diagnostic totem that analyzes a user's skin, lifestyle factors, and behavior patterns, then produces a personalized routine instantly. This creates a retail experience that's data-driven, interactive, and deeply tailored, while also generating insights about consumer needs across stores.

My Role

Product Strategy

Defined the diagnostic flow, mapped user emotions in a retail setting, and scoped the requirements for a fast, trustworthy, in-person experience.

ML Systems & Computer Vision

Implemented the skin-scoring pipeline, tested CV models, designed data features using lifestyle and environmental inputs, and validated results with real users.

Full-Stack Development & Prototyping

Designed and coded the totem interface, built backend logic for scoring + recommendations, and created interactive prototypes to refine usability.

The Solution

An AI-powered in-store experience that:

1

Scores skin concerns using computer vision

2

Interprets lifestyle factors that influence skin health

3

Generates personalized routines with product reasoning

4

Provides a seamless, friendly guided interface

5

Helps retail teams understand customer needs in real-time

Learnings

What I took away

AI works best when the experience feels human.

Building the totem taught me that users trust diagnostics not because they're technical, but because they feel guided, understood, and seen. Designing warmth, clarity, and reassurance into the flow mattered as much as improving model accuracy.

Real personalization requires thinking beyond the model.

I learned that great recommendations come from combining multiple signals: vision, lifestyle, environment, and behavior into one coherent system. True intelligence emerges from the way these pieces interact, not from any single layer.

Check other projects