Skip to content

Case study · Team project

Falcon Eye AI

A browser tool for AI image edits: generative fill, object and background removal, recoloring and restoration, with Stripe credits. Built by four people in 12 days.

My role
Team Lead, Full Stack Developer, AI Engineer, UI/UX Designer
Timeline
Oct 2024, 12 days
Context
Tuwaiq Academy · final project
Platform
Responsive web app with an Express API
Falcon Eye AI landing page and community gallery on two screens
Fig. 01 · Landing page, editor and community gallery

01 Overview

Falcon Eye AI is a web tool for AI image editing. People sign up, upload an image and run an edit: generative fill, object removal, background removal, recoloring or restoration. Edits spend credits bought with Stripe, and a community gallery shows what others made.

  • 12 Days
  • 4 People
  • 5 AI edits

02 Problem

AI image edits are spread across separate apps and sites, and for many people a simple edit still means paying a designer, a cost that adds up. We wanted the common edits in one browser tool, paid per edit, and we had less than two weeks to ship it.

Constraints

12 days from the first commit to launch, four people working in parallel and real payments: every edit had to spend credits that Stripe had confirmed.

03 My role

I led the team, designed the UI/UX and worked on both the client and the server, with the most commits in each repository (78 of 250 and 32 of 81).

  • Landing and brandBuilt the landing page, the about page, the navbar and the footer, and set the logo colors and palette
  • AI editsBuilt generative fill and the image preview, and fixed background removal
  • PaymentsSet up Stripe on the credits page, with the checkout session and webhook on the server
  • APIWrote the request validators for creators, images and transactions
  • AccessSent signed-out visitors to sign-in when they open an edit
Frontend
ReactViteTailwind CSSFramer MotionJavaScript
Backend
Node.jsExpressMongoDB
Services
StripeCloudinaryRender
Tooling
GitGitHub
Design
FigmaCanva

04 Key decisions

The choices that shaped the product. Open one to see why it was made and what it cost.

AI edits through Cloudinary

What we chose

Cloudinary's AI transformations for all 5 edits, instead of hosting our own models

Why

In 12 days, a hosted API let us ship five edits and spend the time on the product around them.

Trade-off

Every edit depends on Cloudinary's models, limits and pricing.

Credits confirmed by Stripe

What we chose

Stripe Checkout on the credits page, and a webhook on the server that records each payment

Why

The server adds credits only after Stripe confirms the payment, so they can't be added from the browser.

Trade-off

A webhook and its keys to set up for each environment: the payment setup took several rounds of fixes in the last three days.

Validate on the server

What we chose

express-validator rules for every creator, image and transaction request

Why

The API checks what the client sends instead of trusting it.

Trade-off

Rules to keep in step with the forms on the client.

05 Result

Shipped on Oct 31, 2024 with sign-up and sign-in, 5 AI edits, Stripe credits, profiles and a community gallery, deployed on Render.

  • 331 Commits across client and server
  • 9 Pages

What I learned

Twelve days taught me to wire the risky part first: payments worked end to end before we polished the screens. Next time I'd test the webhook with Stripe's CLI from day one instead of fixing its setup at the end.

With thanks to

Next project: YC Directory

A startup pitch directory in Next.js and Sanity: sign in with GitHub, submit pitches in Markdown, search them and see view counts.

Questions

Want to Go Deeper on Falcon Eye AI? I’m Happy to Walk You Through It.

Email me