# How I Built ELCV.app in 2 Weeks Using Supabase, Vercel, and Gemini

### A Resume Builder… Built in Just 2 Weeks?

Yes. And here’s how.

[ELCV.app](http://ELCV.app) [is a s](https://elcv.app)imple, multilingual AI-powered res[ume buil](https://elcv.app)der that lets [users cr](https://elcv.app)eate and download a professional CV without needing to sign up. It supports **any language**, includes **AI rewriting tools**, and offers **a clean one-page UX**.

I built it from scratch in just under **three weeks**, solo.

This post is [for make](https://elcv.app)rs, devs, indie hackers, or anyone interested in t[he actua](https://elcv.app)l tech stack and thought process behind a small SaaS product launch.

### Problem First, Code Later

Before I wrote a single line of code, I validated one thing:

> People struggle to write professional CVs — especially in languages other than English.

Most resume tools are:

* English-only
    
* Complex or bloated
    
* Locked behind sign-up or subscriptions
    

So I wrote down 3 core principles:

1. Multilingual input/output
    
2. No sign-up required to generate
    
3. Super cheap pay-as-you-go download model
    

### The Tech Stack

I wanted a serverless, scalable, zero-maintenance setup. Here's what I used:

#### **Frontend**

* **React + TailwindCSS** – lightning fast development and responsiveness
    
* **Next.js on Vercel** – SSR, image optimization, easy deployment
    
* **ShadCN UI** – for modern components
    

#### **Backend / Auth / DB**

* **Supabase**
    
    * Auth: Email/password login (optional)
        
    * Postgres DB: To save resume content (if logged in)
        
    * Storage: For user-uploaded photos and resume backups
        

#### **AI Tools**

* **Gemini API**
    
    * Rewriting user CV content
        
    * Generating cover letters
        
    * Analyzing CV quality (coming soon)
        

#### **Payments**

* **Kashier API (Egypt)** for fast local micro-payments
    
* Stripe or crypto support is planned for global users
    

#### **PDF Generation**

* HTML-to-PDF via `puppeteer-core` on Vercel edge functions
    
* Output options: PDF, JPEG, PNG
    

### Launch MVP, Learn Fast

Instead of waiting for everything to be “perfect”, I launched the first version with:

* 6 templates only
    
* 1 AI prompt per CV
    
* 3 download format (PDF, JPG, PNG)
    
* Only guest mode (no login)
    

The result? People **actually used it**.

### What’s Next?

* More AI features: CV grading, tone correction
    
* More templates: By industry & design
    
* Global payment support (Stripe, Razorpay)
    

### Lessons Learned

* You don’t need to build a giant product to be useful
    
* AI + nocode/devtools = launch fast, iterate faster
    
* Make the *path to value* instant — no forced login, no upsells
    
* A $0.50 product is still a business — if the UX is 🔥
    

### Try ELCV.app

Want to see it in action?  
👉 [https://elcv.app](https://elcv.app)

Try building a resume in your native language, and let me know what you think.

Built solo, shipped fast, and growing every day

Thanks for reading!  
Hossam