www.jetson-ai-lab.com

The NVIDIA Jetson AI Lab is your guide to running generative AI models entirely on-device with NVIDIA Jetson. Explore optimized tutorials, benchmarks, and hands-on examples for LLMs, VLMs, image generation, speech recognition, and more.

Jetson AI Lab pairs a modern Astro frontend with a content-driven workflow, enabling new tutorials, models, and community projects to be published without touching layout code.

• Optimized showcase for flagship models: Llama 4, Gemma 3, Qwen, SDXL, Whisper, and more
• Curated tutorials, benchmarks, and community resources with consistent styling
• Local-first architecture: everything runs on Jetson hardware with no cloud dependency
• Content authored in Markdown/JSON for fast iteration and git-friendly reviews
• Archive of legacy documentation at /archive/

• Astro 5 - Static-first frontend framework
• Tailwind CSS - Design system and styling
• MDX & Markdown - Rich content with interactive elements
• TypeScript - Type safety across data collections
• Chart.js - Interactive benchmark visualizations

/
├── src/
│   ├── components/           Shared UI elements
│   ├── content/              Markdown and JSON content repositories
│   │   ├── home.json         Homepage metrics and featured models
│   │   ├── models/           Model deep dives authored in Markdown
│   │   ├── tutorials/        Long-form tutorials with frontmatter
│   │   ├── projects/         Community project entries
│   │   └── gtc26/            GTC workshop lab content
│   ├── data/                 Benchmark and category data (JSON)
│   ├── layouts/              Base layouts for pages and tutorials
│   └── pages/                Astro routes for the site
├── public/
│   ├── archive/              Legacy MkDocs documentation (static)
│   ├── code-samples/         Downloadable tutorial scripts
│   └── images/               Static assets
├── astro.config.mjs          Astro configuration with redirects
├── tailwind.config.mjs       Tailwind theme definitions
└── TUTORIAL_TEMPLATE.md      Authoring guide for new tutorials

• Node.js 20 and npm
• Git

1. Clone the repository

git clone https://github.com/NVIDIA-AI-IOT/jetson-ai-lab.git
cd jetson-ai-lab

2. Install dependencies

npm install

3. Start the development server

npm run dev

The development server will be available at http://localhost:4321

4. Build for production

npm run build
npm run preview

See CONTRIBUTING.md for collection schemas, validation requirements, and contribution workflow.

1. Create a Markdown file in src/content/tutorials/ with appropriate frontmatter
2. Create a matching .astro wrapper in src/pages/tutorials/
3. Follow the TUTORIAL_TEMPLATE.md guide for formatting

Add a Markdown file under src/content/models/ with frontmatter for overview, benchmarks, and usage examples.

Edit src/content/home.json to update hero metrics, featured models, and stats.

• Use Markdown headings and tables for consistent styling
• Match tutorial frontmatter to src/content/config.ts (title, description, category, tags, and optional authors)
• Store media assets under public/ and reference with absolute paths

Old MkDocs URLs are automatically redirected:

Redirects are configured in astro.config.mjs and the custom 404.astro page.

The site is automatically deployed to GitHub Pages via GitHub Actions on every push to main.

• Production URL: https://www.jetson-ai-lab.com
• Build Output: Static HTML/CSS/JS in dist/

The .github/workflows/ci.yml workflow handles:

• ✅ Automated builds on push to main
• ✅ Deployment to GitHub Pages
• ✅ Node.js 20 with npm caching

Legacy MkDocs documentation is preserved at /archive/ with a deprecation banner linking to the new site. This ensures old bookmarks and external links continue to work.

Developers inside and outside NVIDIA are invited to contribute tutorials, Jetson projects, model updates, code samples, benchmarks, and site improvements. Open your pull request against main, and one of our core contributors will review it.

See CONTRIBUTING.md for setup, publishing guidelines, validation requirements, and the pull request process.

Licensed under the MIT License.

For questions or contributions, please open an issue or contact the maintainers.

• Khalil Ben Khaled
• Aditya Sahu
• Chitoku Yato