Inspiration

My inspiration comes from watching people and celebrities who would spend thousands of dollars to go to Korea just to do their color analysis, face and body recommendations, makeup tips ,etc. And I believe getting dressed shouldn't require a personal stylist's fee. Most people default to the same few outfits, makeup, colors or style that wont even suit them because figuring out what actually suits their face shape, color season, and style is hard without expert guidance. I wanted to put that expertise in everyone's pocket powered by AI, not a subscription to a human stylist most people can't afford.

What it does

Fashionist is an AI powered personal styling app. Using the Gemini API, it analyzes a user's face shape, color season, and style preferences to generate personalized outfit, accessory, and makeup recommendations down to specific details like frame shapes for glasses or flattering lipstick shades.

Every recommendation comes with a "shop this" icon that opens a real, purchasable product search on Amazon through an integrated affiliate link turning AI inspiration into something users can actually buy, instantly.

Beyond styling, Fashionist is a full social platform: users can post, follow other real users, discover people through a dedicated "Suggested for You" feed, and message each other.

How we built it

It was build using :- Gemini API powers the core recommendation engine for face shape/color analysis, styling logic, and generating structured outfit and shopping-search data. Firebase Realtime Database stores users, posts, and social data. Express.js backend, deployed on Vercel, connected to GitHub for continuous deployment. Amazon Associates affiliate integration turns each AI recommendation into a shoppable link. Dodo Payments handles the Premium subscription checkout , chosen specifically because it supports individual developers accepting global payments without requiring a registered business entity.

Challenges we ran into

Navigating payment provider restrictions as an not a registered company most major processors (Stripe-backed platforms) don't support individuals in India accepting live international payments, which led to switching payment providers mid-build. Debugging Firebase Realtime Database security rules to safely expose user discovery data without leaving the entire database publicly writable.

Accomplishments that we're proud of

Shipping a fully working, live, deployed app with real AI driven recommendations, real affiliate monetization, and a real payment-gated premium tier as even a late build.

What we learned

Building Fashionist end-to-end surfaced lessons well beyond just writing code:

AI output needs guardrails, not just prompts. Getting Gemini to return genuinely useful, specific data (like a real shopping search term instead of a vague color name) took iteration the first version technically worked but produced results that made no sense to a real user. Payment infrastructure isn't one-size-fits-all globally. As an individual developer based in India, most mainstream payment platforms (Stripe and Stripe-backed tools like Lemon Squeezy) simply don't support individuals accepting live international payments a constraint that isn't obvious until you hit it, and it reshaped a core part of the build. Security and functionality are often in tension, not opposition. Getting real user discovery to work in Firebase meant resisting the easy fix (making the whole database public) in favor of scoping access precisely to what the feature actually needed.

What's next for Fashionist

Smarter discovery and recommendations ranking "Suggested for You" by shared style interests or mutual connections rather than a random list of users. Expand AI styling categories beyond outfits, glasses, and makeup into wardrobe planning, etc. Deepen the affiliate layer supporting more retailers beyond Amazon, and refining search-term generation further for even more accurate product matches. Community features style challenges, and creator-style profiles to make Fashionist feel like a genuine fashion social platform, not just a recommendation tool.

Built With

Share this project:

Updates