Let's not romanticize this.
I don't spend weeks "engineering" these projects from scratch.
Most of the time the workflow is:
- Get one of my old blog post
- Extract the hook (the metaphor / mental model / sharp point)
- Ask Google AI Studio to build it as a TypeScript webapp
- Refine the UI / UX with prompts until it matches the hook
- Push to GitHub and let Copilot publish it via Github pages
That's it. All done with my smartphone.
The trick is having a hook that is concrete enough to survive translation into an interaction.
Once you have that, "blog post → webapp" becomes a simple task:
I own the Why (the hook).
AI produces the How (the first draft).
1. Code Quality & Maintenance
Projects:
- CodeFengShui — Chaos → clarity slider (blog post / code)
- CodeHygiene — Interactive teaching tool (blog post / code)
- CodebaseKitchen — Entropy resistance test (blog post / code)
- TheInnerCodePlumber — Systemic intuition model (blog post / code)
These simulate entropy over time.
They exist to demonstrate a brutal truth:
Code doesn't break suddenly.
It decays quietly.
You don't refactor because of a bug.
You refactor because the system feels wrong.
These tools train that feeling.
2. Software Principles & Learning
Projects:
- SOLID_Paw — SOLID via Paw Patrol (blog post / code)
- LegacyPHP — Save the Frankenstein monster (blog post / code)
- ChroniclesCleanCode — Clean code superhero arc (code)
- learnings — Personal knowledge base (blog post / code)
These are not tutorials.
They are memory anchors.
Nobody remembers principles because of definitions.
They remember them because of stories, metaphors, and failure modes.
Cringe works.
That's the point.
3. AI / LLM & Prompt Engineering
Projects:
- BlindSpotMetaPromptApp — Force problem clarity (blog post / code)
- PromptMastery — Prompt techniques (blog post / code)
- PromptDoc — Documentation prompts (code)
-
your_agent_config — Build
AGENTS.md(code) - TheAntiCodingAgentGuide — Satirical survival guide (blog post / code)
These don't help you "use AI better".
They help you not lie to yourself.
AI doesn't fail because it's stupid.
It fails because humans skip the thinking step.
Yes, I use AI to build tools about using AI carefully.
That irony is intentional.
4. PHP & Systems Development
Projects:
- PHPStanGuide — Type system via UI (blog post / code)
- PHP_Stack_Timeout_Simulator — Timeout visualization (blog post / code)
- MagicTypeCube — PHP wisdom cubes (blog post / code)
Backend failures are invisible by default.
These projects make constraints visually unavoidable.
If you can see it, you stop arguing about it.
5. Career, Industry & Culture
Projects:
- TheStateOfSoftware2026 — Industry analysis (blog post / code)
- STRANGER_IT_THINGS — Incident vs request game (code)
- MentaleDeveloperMap — Mental model navigation (code)
- learnings — What I've learned so far (blog post / code)
Software is not just technical.
It's economic, cultural, and political — whether we like it or not.
Ignoring this layer doesn't make it disappear.
It just makes you unprepared.
Happy Coding. ❤️
Top comments (0)