🎯 What This Is
A 30-day AI System Design curriculum built for someone who learns by building. Each day is calibrated to 30 minutes: 15 minutes of concept + 15 minutes of hands-on, runnable on real infrastructure (your VPS, Docker, llama.cpp, whatever stack you choose).
This isn't a book course — it's a living portal. Content is written in Markdown, served by a Flask app in Docker, and designed so that different Hermes agent profiles can collaborate on it.
🤖 How The Content Gets Made
Ayva (researcher)
Researches each day's topic — digs into papers, docs, and real-world case studies. Produces detailed theory content, references, and exercise design.
Codi (implementer)
Takes Ayva's research and builds the hands-on exercises — containers, scripts, benchmarks, mini-projects that run on real infra.
You (learner)
30 minutes a day. Read the theory, run the exercise. The portal tracks what's ready and what's being worked on.
🏗 Architecture
Everything runs in a single Docker container. Content lives in
/content/week-*/day-*.md — add or edit a markdown file
and the portal picks it up on next refresh (no rebuild needed).
📁 Content Format
Each day file uses YAML frontmatter + Markdown body:
---
title: Your Title
topic: Short description
status: ready # or "stub"
category: Foundations
day: 1
week: 1
---
## Learning Objectives
...
## Theory (15 min)
...
## Hands-on (15 min)
...
## Key Takeaways
...⚙️ Workflow
- Ayva researches a topic and writes/updates a
day-NN.mdfile - Codi reviews, builds the hands-on exercise, and commits the implementation
- Portal reflects the updated status badge automatically
- You open the day, read, run the exercise
📊 Stats
| Total days | 30 |
| Minutes per day | 30 |
| Total hours | 15 |
| Weeks | 4 |
| Status | Ready + Stub |