INSPIRATION

I built Peach Match for Anastasia. She inspired it, and she is my first tester.

80% of women of color say they can't find a foundation that matches their skin. The industry answered with 50-shade ranges, and it still doesn't work — because the problem was never the number of shades. It was undertone. A shade can be the right depth and still look ashy on you if the undertone is wrong. That is the problem this app fixes, with one selfie.

WHAT IT DOES

Snap a selfie and the app guides you to take a usable photo first — face the window, no flash, hair back — because bad lighting is why most shade apps fail.

Then YouCam AI reads your exact skin tone, your Fitzpatrick type, and skin scores for radiance, moisture, clarity and oil balance.

Our own matcher compares your real skin tone against 78 foundation shades from 9 brands in CIELAB, the color space cosmetics labs use, and ranks them with your undertone weighted in — so a shade that flatters your undertone outranks a marginally closer depth that would look ashy.

Then you see the shade on your own face with virtual try-on, and the app shows its working: your undertone, why the depth matches, and what the shade actually looks like on you. Tap any alternate shade and the try-on re-renders.

Also included:

  • Shade Passport that saves your match
  • Concealer finder
  • Makeup bag audit that tests foundations you already own against your skin tone

HOW WE BUILT IT:

React Native with Expo SDK 57 and TypeScript, using Expo Router.

The pipeline uses YouCam APIs for file upload, skin tone reading, Fitzpatrick scale, skin analysis, and makeup virtual try-on. The matching itself is my own code — CIELAB color distance plus hue-angle undertone analysis.

The catalog is 78 shades weighted toward deep and dark tones, from Fenty, MAC, Maybelline, Black Opal, NARS, Lancome, L'Oreal, Estee Lauder and Rare Beauty.

CHALLENGES WE CAME ACROSS

The first one was my own mistake: I mixed up the API Key and the Secret Key, so every call came back 401 until Perfect Corp's support pointed it out.

Then the upload itself — the presigned PUT has to match the ticket headers exactly, or it breaks.

The harder problem was honesty: bad lighting produces bad matches, so instead of guessing, the app refuses to guess on a bad photo and tells you to retake. And for the try-on, skin smoothing is set to zero, so the render shows the actual shade instead of a beautified blur.

ACCOMPLISHMENTS THAT WE ARE PROUD OF:

The live YouCam pipeline works end-to-end on a real phone, not just in demo mode. The undertone-weighted matcher, the draggable before/after try-on slider, and the Shade Passport all shipped, and the app stays honest on a bad photo instead of giving a confident wrong match.

WHAT WE LEARNT:

Undertone matters more than shade count. Color science is not abstract — CIELAB distance and hue angle are what make a match look right on a real face. And never guess on a bad photo: one honest retake prompt is worth more than a confident wrong answer.

WHATS NEXT FOR PEACH MATCH

The Peach Cosmetics storefront inside the app is COMING SOON — the store is still being built, so right now it is pre-order and waitlist, not live ordering. Next is more brands in the catalog and an iOS build.

Built With

Share this project:

Updates

Submission history