Your outfit, scored by AI. Know exactly what's working — and what to fix.
Dripmax is an iOS-first React Native app that analyses full-length outfit photos with Google Gemini. You upload a mirror selfie; you get a 0–100 score across four style axes, a named style archetype, and a coaching report that tells you why the look lands — and exactly what to change.
Built solo by Sanda ANDRIA at exeio Corp.
This repository is the public showcase. The application source is private. See Why the source is private.
A user photographs their outfit. Gemini returns a structured analysis in under 15 seconds:
| Output | Description |
|---|---|
| Drip Score | 0–100 overall, with a labelled band (Editorial → Incoherent) |
| Archetype | a named style identity — "The Minimalist", "The Streetwear Maven" |
| Four Axes | Fit & Silhouette · Colour Harmony · Cohesion & Intent · Details & Finishing |
| Free read | one honest line per axis |
| Priority Fixes (Pro) | ranked fixes with impact levels — HIGH / MED / LOW |
| Quick Styling Moves (Pro) | three things to do right now, without buying anything |
| Full Report (Pro) | deep axis diagnosis, shopping direction, trend tie-in |
The reveal screen surfaces the score immediately. The full report is the deep dive.
Style is subjective, and a model asked "is this outfit good?" produces confident, useless generalities. Getting a reproducible 0–100 score out of a vision model means the subjectivity has to be pushed somewhere it can be constrained.
The approach: decompose the judgement into four independent axes with their own rubrics, score each one separately, and derive the overall number from them. The model never scores "the outfit" — it scores fit, then colour, then cohesion, then finishing. The composite is arithmetic.
That single decision is what makes the output stable enough to charge for, and it's the part that took the longest to get right.
Mobile — React Native · Expo · TypeScript · Zustand AI — Google Gemini (vision) Backend — Firebase Payments — RevenueCat (free / Pro tiers) Platform — iOS-first
Dripmax is a commercial product with a paid tier. The prompt engineering and rubric design are the product — they're the result of a long iteration loop, and publishing them would mean publishing the thing people pay for.
What is public: how it's built, what it does, and what I learned building it. I write about that in detail — see below.
I write about building consumer AI products solo while holding a full-time data engineering job — architecture decisions, what broke, and the numbers.
- Newsletter — (launching soon)
- exeio Corp — exeiocorp.com
- GitHub — @ExeioS33
exeio Corp builds consumer apps for people with taste. Technology should feel like it was made for you, not sold to you.