Skip to main content
Cloudflare Workers · Cloudflare R2 · JavaScript · Serverless · Wrangler

​
Overview

A serverless file upload and management system using Cloudflare Workers as the compute layer and Cloudflare R2 as object storage. There is no file size limit by default, making it well suited to machine learning model weights (.pt, .ckpt, .safetensors, etc.). The frontend can be hosted on Cloudflare Pages, GitHub Pages, or opened locally as a static HTML file. The backend Worker is deployed and managed with the wrangler CLI.

​
Interface preview

Click “Simulate upload” to feel the real drag-and-drop flow (this is a frontend mock and does not upload anything).

​
Core features

  • No file size limit; handles large model files without configuration changes
  • Drag-and-drop upload interface with real-time progress and estimated time remaining
  • Automatic file-type detection with matching icons
  • Image preview and download for all file types
  • File deletion from the management interface
  • Cloudflare Workers handle upload, download, and delete API requests
  • Flexible frontend hosting: Cloudflare Pages, GitHub Pages, or local static file

​
Architecture

Frontend and backend are fully decoupled. The Worker is deployed to Cloudflare’s edge with wrangler deploy; the frontend only needs the correct Worker URL and can be hosted anywhere.

​
Quick start

1

Install wrangler CLI and log in

2

Create an R2 bucket

3

Configure wrangler.toml

Fill in the Worker name, R2 bucket binding, and bucket name:
4

Update the frontend API URL

In frontend/app.js, set API_URL to your Worker endpoint:
5

Deploy the Worker

6

Deploy or open the frontend

Upload the frontend/ directory to Cloudflare Pages or GitHub Pages, or open frontend/index.html directly in a browser.

​
Notes

CORS is set to * by default and there is no authentication. Anyone who knows the Worker URL can upload, download, or delete files. Before using this in production, add an authentication layer and restrict CORS to specific origins. Without these changes the storage is effectively public. R2 usage may also incur Cloudflare charges; monitor your usage limits.

​
In practice

A practical fit for individuals or small teams who need a place to store large binary files (model weights, dataset archives) without running a backend server. It also works as a hands-on starting point for learning Cloudflare Workers and R2 together.