TikTok’s days may be numbered, so I'm here to save the day! TikTok is just an infinite scrolling video feed, a smart recommendation algorithm, and some cloud servers. In this video, I challenge myself to vibe code the frontend and backend for a fully functional TikTok clone from scratch using React Native, Django, AWS, and AI tools like GPT-4 and Copilot in under 24 hours. And if I fail... I post an embarrassing TikTok.
//App Information
PopOff has been removed from the app store.
The app was available to download in US & Canada from 05/23/25 to 07/24/25
//Tech Stack
Frontend: React Native, Expo, JavaScript
Backend: Django, REST API, PostgreSQL, Amazon RDS, Amazon S3, Amazon EC2, Cloudfront, Celery, Redis, GPT-4, Whisper
The project has been made open source here:
https://github.com/Carson-Stark/PopOffFrontend//Socials
Instagram: tesseract.yt
TikTok: tesseractyt
//Chapters
00:00 - Intro
01:05 - Feed Template
02:58 - Backend Explanation
05:52 - Uploading Videos
08:05 - Issues
09:10 - Post Records
10:35 - Profile, Likes, Comments
12:25 - Recommendation Algorithm
15:10 - End of Time
17:13 - Recommendation Improvements
18:33 - New Video Embeddings
21:43 - Testing Embeddings
23:31 - Launch / Final Demo
24:40 - Now What?
25:39 - Vibe Coding
//Music
Context Sensitive - Virtually Impossible
https://www.youtube.com/ContextSensitivehttps://contextsensitive.bandcamp.com/Epidemic Sound
//Long Description
When TikTok was temporarily banned in the U.S., it raised a simple but powerful idea: what if I could rebuild TikTok’s core functionality in just 24 hours? In this video, I take you step-by-step through building a full-stack TikTok clone with a vertically scrolling video feed, real-time video playback, user profiles, likes, comments, and post uploads. The app is built using React Native and Expo on the front end, and Django with Django REST Framework on the backend. I use PostgreSQL for relational data, AWS S3 for video storage, and EC2, RDS, and CloudFront for deployment and delivery.
You’ll see how I implement media uploads using pre-signed S3 URLs, create API endpoints to handle content and user actions, and dynamically populate the feed with real user data. I also develop a recommendation system using OpenAI’s Whisper to transcribe audio, GPT-4 to summarize video content, and embeddings to rank videos by user interest. The system updates in real time as users interact with content. Background tasks are handled with Celery, and videos are optimized using HLS streaming. Along the way, I reflect on how AI tools like ChatGPT and GitHub Copilot accelerated the build—and what it means for indie developers building production-grade apps solo.
//Hashtags
#mobileappdevelopment #reactnative #django #python #tiktok #coding #24hourchallenge #fullstackdeveloper #webdevelopment #vibecoding