A living logbook of applied-AI practices.
A bilingual (ES / EN) curriculum + a Radar feed that stays current. Not a course that ages — when something new ships in applied AI, it lands on the Radar and the module it touches gets updated.
Built and maintained by the Dmeter team · open to every dev 💚
Good applied-AI material ages fast, and almost all of it is in English. DevLog is the opposite: a living, bilingual logbook built to be read and to stay current. It's a school (the curriculum) and a logbook (the Radar) at once — a shared notebook for every dev shipping AI in the real world.
- Curriculum — 12 modules, Fundamentals → Production. Every module follows the same didactic format: why it matters → diagram (Mermaid) → how it works → code (Python + TypeScript) → mini-lab → mastery checklist → common mistakes → when NOT to use it → sources.
- Radar (Logbook) — a chronological feed of what changed in applied AI, why it matters, and which module it affects. The reason to come back each month.
- Authors & contribution — a module template and guide so the community can add pieces with the same rigor and format.
The syllabus reflects the 2026 market: Context Engineering as the parent discipline, modern RAG (vector + hybrid + reranking), Agentic RAG, GraphRAG, multitenancy & security, agents with MCP, multi-agent with A2A, agent memory, evals/observability/guardrails, long-context, and an end-to-end capstone.
| # | Module | Level |
|---|---|---|
| 00 | The 2026 landscape | Fundamentals |
| 01 | Context Engineering | Fundamentals |
| 02 | Modern RAG | Fundamentals |
| 03 | Agentic RAG | Intermediate |
| 04 | GraphRAG | Intermediate |
| 05 | Multitenancy & Security | Production |
| 06 | Agents & MCP | Intermediate |
| 07 | Multi-agent & A2A | Production |
| 08 | Agent memory | Intermediate |
| 09 | Evals, Observability & Guardrails | Production |
| 10 | Long-context & when NOT to use RAG | Intermediate |
| 11 | Capstone: a multitenant agentic SaaS | Production |
- Astro + Starlight — documentation site
- astro-mermaid — diagrams written as text (
```mermaid) - Bilingual i18n (
src/content/docs/esandsrc/content/docs/en) - Custom Swiss / monochrome theme, custom theme & language dropdowns
npm install
npm run dev # http://localhost:4321
npm run build # output in dist/
npm run preview # preview the builddevlog/
├─ astro.config.mjs # site config · i18n · mermaid · component overrides
├─ src/
│ ├─ content.config.ts # Starlight docs collection
│ ├─ styles/theme.css # the design system
│ ├─ components/ # custom Footer, ThemeSelect, LanguageSelect
│ └─ content/docs/
│ ├─ es/ # Spanish content
│ │ ├─ index.mdx # homepage
│ │ ├─ curriculum/ # 00..11 modules
│ │ ├─ logbook/ # Radar (the living feed)
│ │ ├─ contribute/ # guide + module template
│ │ └─ authors/ # credits + how to join
│ └─ en/ # same tree, in English
DevLog is open and it grows with the community — PRs are welcome. Whether you fix a typo, sharpen an explanation, add a Radar entry, or write a whole new module, you're in.
- Read the contribution guide and the module template under
src/content/docs/{es,en}/contribute. - Pick a new module, an improvement to an existing one, or a Radar entry.
- Keep content bilingual (ES + EN) and follow the fixed format.
- Run
npm run buildbefore opening your PR (it validates links and Mermaid). - Open the PR — once it's merged, you're added to the Authors page.
The quality bar is high on purpose: clarity, engineering honesty, and the fixed format. That's what makes DevLog feel like one product instead of scattered notes.
The Radar is what makes this a product instead of a static course. Suggested cadence: a
monthly pass (or same-day when something big drops). Each entry is short —
what shipped → why it matters → which module it affects. See
src/content/docs/es/logbook and the contribution guide.
Dmeter builds custom software and AI products. DevLog is our open educational arm — a living logbook we keep current so the whole dev community can learn applied AI the way the market actually builds it.
Content: CC BY 4.0 · Example code: MIT.
Built and maintained by the Dmeter team. Issues and PRs welcome 💚