
The short answer
Ewoi Builds is a Laravel 13 application with a MariaDB database, designed page by page before any code, with security and speed decided up front rather than added later. This log explains the main decisions and the trade-offs behind them, so you can reuse the reasoning in your own projects.
Build log, 14 to 15 November 2026.

A journey is not a tutorial. It is the honest story of a build: what we chose, why, and what it cost. Read it for the thinking.
The goal
A free place to learn software engineering, built for learners in Kenya first and everyone else too. It had to:
- Load fast on a phone over a slow connection.
- Teach complete beginners without assuming anything.
- Keep learners' data safe, with no shortcuts on sign-in or passwords.
- Run on affordable hosting, without a team of operations engineers.
Every decision below comes back to one of those four.
Decision 1: design every page before writing code
Before the first line of Laravel, every page existed as a design: public pages, the learner dashboard and every email. Each design was approved, and a tool compares the built page with the approved screenshot on every change. If the build drifts from the approved design, the check fails before anything ships.
Why: changing a design is cheap; changing a built page is expensive. The comparison also catches accidents, such as a style change that breaks a page nobody was looking at.
Trade-off: slower to start. Worth it: almost no page needed rework after it was built.
Decision 2: a boring, well-known stack
| Part | Choice | Why |
|---|---|---|
| Back end | Laravel 13 on PHP | Mature, secure defaults, runs on common hosting |
| Database | MariaDB | Fast, free, supported by most hosts |
| Front end | Blade templates with a little Alpine.js | Pages are rendered on the server, so they are fast and readable by search engines |
Why: a stack many developers know is easier to hire for, debug and host. New and exciting tools can wait until there is a real problem they solve.
Decision 3: UUIDs instead of counting IDs
Most tutorials give every row an ID that counts up: 1, 2, 3. We used UUIDs, long random identifiers such as 0196f1c4-..., for every table.
Why: counting IDs leak information (how many users you have) and invite guessing (/invoices/41, then /invoices/42). Random IDs cannot be guessed or counted, and an automated check keeps it that way.
Trade-off: UUIDs take more space and are harder to read in logs. MariaDB has a native uuid column type, which keeps them compact.
Decision 4: security from the first commit
Passwords are never stored as readable text (why that matters), every input is validated, and protection against common attacks was part of the first version rather than a later task.
Why: security added later is always partial. Starting secure costs a little more on day one and much less afterwards.
Decision 5: fast pages for slow connections
- Public pages are cached where it is safe, so most visits never touch the database.
- Fonts are hosted with the site instead of loaded from another service.
- Images are WebP, sized for their slot, and loaded lazily below the fold.
- Page views are counted with a tiny request after the page loads, so cached pages still count, and without cookies.
Decision 6: teach beginners properly
Every lesson defines its words (the glossary links), has a copy button on every command, an AI prompt for when you are stuck, Ewoi's tips, a quiz and a request box. A test checks every lesson for these, and for broken links, before anything ships.
What we would do differently
- Write the content library earlier. Building pages around placeholder text hid layout problems that real, longer text revealed.
- Decide the booking rules (hours, notice, limits) before designing the training page, not after.
Ask your AI assistant
Here is the list of technical decisions for my own project: [paste them]. For each one, ask me what problem it solves and what it costs, then tell me which decision I have not justified well.
Key takeaways
- Decide what matters (speed, safety, beginners, cost) and test every choice against it.
- Design before you build, and check the build against the design automatically.
- Prefer well-known tools; choose new ones only for a real problem.
- Make security and speed defaults, not features.
Check your understanding

Every lesson stays free. Tap the thumbs up so we know what works, or say thanks with a soda or a coffee: it pays for hosting and new lessons.



