Lossless Network

AI networking, distilled — for network engineers entering the AI era.

A public learning hub that bridges network engineers into AI fluency. 21 modules, 85 hands-on labs, 6 break-fix scenarios, written for engineers who already know BGP, OSPF, ECMP, and fabric design — and now have to support GPU clusters that look nothing like a CLOS data center.

Brand promise: Deep when you need depth. Fast when you need speed.

---

Repo layout

losslessnetwork/
├── docs/             ← Curriculum (markdown) — all 21 modules + intro
│   └── modules/      ← Phase 0 to Phase 6
├── src/              ← React components (LabActions, Homepage)
├── static/           ← Logos, hero image, asciinema casts
│   ├── img/
│   └── casts/        ← .cast files for in-browser terminal demos
├── blog/             ← Blog posts, RSS, tags
├── lab/              ← Container-based GPU fabric simulator
│   ├── docker-compose.yml
│   ├── launch.sh / start.sh / stop.sh / teardown.sh
│   ├── infra/        ← KIND, containerlab, k8s manifests, nginx
│   ├── images/       ← Mock containers (worker, mock-switch, mini-trainer, etc.)
│   ├── scenarios/    ← 6 break-fix exercises
│   ├── simulator/    ← Python simulator (engine.py, run.py)
│   ├── assessment/   ← Knowledge checks
│   └── cert/         ← Completion certificate templates
├── docusaurus.config.ts
├── sidebars.ts
├── package.json
└── README.md

---

Run locally

Just the website (read the modules)

export PATH="/opt/homebrew/opt/node@20/bin:$PATH"   # macOS: Node 20+ required
npm install
npm start -- --host 127.0.0.1 --port 3000
# Open http://127.0.0.1:3000

The interactive lab (run real GPU fabric simulation)

cd lab
docker compose up -d
# Open http://localhost:8080  (tutorial + browser terminals)

Only prerequisite: Docker. launch.sh handles KIND, kubectl, ttyd, and image builds.

The lab in your browser (zero install)

Click any Open in Codespaces button in the modules — GitHub gives every user 60 free hours/month.

---

Release cadence

One module polished and released every 1–2 weeks. Released modules carry status: published in their frontmatter. Drafts are visible but marked as work-in-progress.

See CHANGELOG.md for release history.

---

License

• Code (Docusaurus site, lab containers, simulator, scripts): Apache 2.0
• Written content (modules, blog posts, docs): CC BY 4.0

Share, remix, build on it. Just credit the source.

---

Author

Built by a Staff Network Engineer who got tired of vendor decks and wrote the playbook he wished existed when he started bridging from traditional DC fabrics to AI training networks.

Personal views. Not affiliated with any employer.