🤖 Gemini YouTube Automation

A fully autonomous AI bot that writes, produces, and uploads YouTube lessons daily — zero human input required.

Demo

What It Does

Every day at 7:00 AM UTC, this bot runs entirely on GitHub Actions and:

Reads the content plan to pick the next pending lesson
Writes a full multi-slide script using Gemini 2.5 Flash
Generates narration audio (gTTS) and fetches Pexels background imagery
Renders a professional slide-based video (1920×1080) with background music
Renders a vertical YouTube Short (1080×1920) from the same lesson
Creates a custom thumbnail for each format
Uploads both videos to YouTube with titles, descriptions, and hashtags
Updates content_plan.json and commits it back to the repo

No local machine. No manual steps. Every video in the channel was made by this pipeline.

How It Works

GitHub Actions Scheduler (7 AM UTC)
          │
          ▼
  ┌───────────────────┐
  │  content_plan.json │  ◄── picks next "pending" lesson
  └────────┬──────────┘
           │
           ▼
  ┌─────────────────────────────────┐
  │  Gemini 2.5 Flash               │
  │  • 7–8 slide lesson script      │
  │  • 1-sentence YouTube Short     │
  │  • hashtags + metadata          │
  └────────┬────────────────────────┘
           │
           ▼
  ┌──────────────────────────────────────┐
  │  Video Renderer (MoviePy + PIL)      │
  │  • gTTS narration per slide          │
  │  • Pexels background images          │
  │  • Background music mix              │
  │  • Long-form (16:9) + Short (9:16)   │
  └────────┬─────────────────────────────┘
           │
           ▼
  ┌────────────────────────┐
  │  YouTube Data API v3   │  ◄── uploads with thumbnails
  └────────┬───────────────┘
           │
           ▼
  ┌────────────────────────┐
  │  git commit + push     │  ◄── marks lesson "complete"
  └────────────────────────┘

Features

Zero-touch operation — fully autonomous, runs on a cron schedule
Dual-format output — long-form lesson video AND a YouTube Short per day
AI-generated curriculum — Gemini creates and extends the course plan automatically
Dynamic visuals — Pexels stock imagery matched to each slide topic
Professional audio — per-slide narration with soft background music
Custom thumbnails — auto-generated for every video
Self-updating repo — content plan committed back after each successful run
GitHub Actions native — no server, no hosting cost, just free CI runners

Tech Stack

Component	Technology
AI Script Generation	Google Gemini 2.5 Flash
Text-to-Speech	gTTS
Video Rendering	MoviePy + FFmpeg
Image Generation	Pillow (PIL) + ImageMagick
Stock Footage	Pexels API
YouTube Upload	YouTube Data API v3
Automation	GitHub Actions

Setup

1. Clone the repo

git clone https://github.com/ChaitanyaEswarRajeshJakki/gemini-youtube-automation.git
cd gemini-youtube-automation

2. Install dependencies

pip install -r requirements.txt

3. Configure GitHub Secrets

Go to Settings → Secrets and variables → Actions and add:

Secret	Description
`GOOGLE_API_KEY`	Gemini API key from Google AI Studio
`PEXELS_API_KEY`	Pexels API key
`CLIENT_SECRET_B64`	YouTube OAuth `client_secrets.json` encoded in base64
`CREDENTIALS_B64`	YouTube OAuth `credentials.json` encoded in base64

4. Run locally

python main.py

Content Progress

The bot is currently producing the "AI for Developers" series — a beginner-friendly course that takes developers from zero to advanced AI.

Topics covered include: Generative AI, LLMs, Prompt Engineering, RAG, Vector Databases, LangGraph, Fine-tuning, Computer Vision, and more.

Track progress live in content_plan.json.

Daily Production Infographic

Star History

Contributing

Contributions are welcome. Open an issue or submit a pull request.

License

MIT License. See LICENSE for details.

gemini-youtube-automation