This repository showcases a full-stack engineering project for a YouTube video generation SaaS. It uses TypeScript, Python, CSS, PLpgSQL, and JavaScript, focusing on architecture and key engineering decisions for production-grade applications.
From the README
🎬 VideoForge — AI-Powered YouTube Video Generation SaaS Script → Audio → B-roll → Final MP4, fully automated. Built to demonstrate production-grade full-stack engineering for US remote job applications. --- 📐 Architecture Overview --- 🚀 Key Engineering Decisions (For Interviewers) Concern Decision Why --------- LLM inference Groq (LLaMA 3.1 8B) Free tier, 500 tokens/sec, no GPU needed on server TTS Gemini 2.5 Flash + edge-tts fallback Multi-key round-robin; zero cost; fallback = 100% uptime Task queue Celery + Upstash Redis Decouples API from heavy work; retry logic; step-level logging Auth Supabase JWT Handles social login, email, MFA; RLS enforces data isolation Video storage...
Skills
- TypeScript
- Python
- CSS
- PLpgSQL
- JavaScript
