Engineering Lab
See how the software behind a project actually works
Six small interactive explainers for people who commission software and want to understand what they are paying for. Each one is a labelled simulation: nothing here calls a real server, measures a real site or shows real output.
API request flow
SimulationA request passes several layers, and each can say no. Knowing where it stopped tells you what to fix.
- BrowserSends GET /api/projects/42 with an Authorization header.
- Edge and rate limitTerminates TLS and checks the caller has not exceeded the request limit.
- AuthenticationVerifies the token signature and expiry to learn who is calling.
- HandlerValidates the input, checks the caller may read this project, then asks the database.
- DatabaseReturns the row for project 42 (or fails).
- ResponseSerialises the result and sends it back with a status code.
Step through to see what each layer does and where a request can stop.
Simulation: scripted steps, no network request is made.
Authentication and authorisation
SimulationAuthentication asks who you are. Authorisation asks what you may do. They fail differently, with 401 and 403.
403 Forbidden
Authorisation failed: the server knows who you are, and your role does not include this action.
| Action | Not signed in | Member | Owner |
|---|---|---|---|
| View the dashboard | 401 | Yes | Yes |
| Create a feedback link | 401 | Yes | Yes |
| Read the access log | 401 | 403 | Yes |
| Delete a record | 401 | 403 | Yes |
The check always runs on the server. Hiding a button in the page is a courtesy, not security: anyone can send the request by hand.
Simulation: a small rule table modelled on a typical team area, not a live login.
Database relationships
SimulationData is split into tables that point at each other. Follow one customer through orders and products.
customers
- 1 · Adalinked
- 2 · Tunde
- 3 · Mei
orders (customer_id)
- #101 → customer 1linked
- #102 → customer 1linked
- #103 → customer 2
- #104 → customer 3
- #105 → customer 3
order_items (order_id, product_id)
- #101 · product 1 × 2linked
- #101 · product 2 × 1linked
- #102 · product 3 × 1linked
- #103 · product 1 × 1
- #104 · product 4 × 1
- #105 · product 1 × 3
- #105 · product 3 × 1
products
- 1 · Notebooklinked
- 2 · Pen setlinked
- 3 · Desk lamplinked
- 4 · Backpack
Ada has 2 orders and bought 3 different products: Notebook, Pen set, Desk lamp.
SELECT DISTINCT p.name FROM customers c JOIN orders o ON o.customer_id = c.id JOIN order_items i ON i.order_id = o.id JOIN products p ON p.id = i.product_id WHERE c.id = 1;
One-to-many: a customer has many orders. Many-to-many: orders and products need the join table order_items. Storing each fact once is what stops a product rename from having to be fixed in a hundred places.
Simulation: a tiny in-page dataset, not a real database.
Responsive layouts
SimulationOne codebase rearranges itself for each screen. Drag the width to watch the rules change.
Headline that wraps to fit
Text keeps a readable line length at every width.
At 390px: 1 card column and a menu button in place of the links. The preview is capped to the space available on your screen.
Real sites are tested at actual widths from 320px up, on real touch screens, with text sizing and zoom changed. This sketch only shows the idea of layout rules changing at breakpoints.
Simulation: a mock layout, not a screenshot of a real site.
Page performance
SimulationMost slow pages are heavy pages. See how size and connection speed combine, and what the usual fixes buy you.
Estimated time to load this model page
3.5 s
3600 KB transferred
- Images2400 KB
- JavaScript900 KB
- Fonts180 KB
- Other120 KB
This is a simple model, not a measurement: time = four round trips + size ÷ bandwidth, starting from assumed page weights. Real pages are measured with tools such as Lighthouse and real-user data.
Simulation: a simple model with assumed page weights. It is an illustration, not a benchmark.
Cloud architecture on AWS
SimulationDifferent workloads need different building blocks. Choose one to see a common set and why each part is there.
Users sign in and the app reads and writes data.
Part 1
CloudFront
CDN in front of everything
Caches static assets and absorbs traffic spikes.
Part 2
Containers (ECS Fargate) or Lambda
Runs the application code
Scales with load without managing machines.
Part 3
RDS (PostgreSQL)
Relational database
Transactions and relationships, with managed backups.
Part 4
S3
User uploads and files
Keeps large files out of the database.
Part 5
Secrets Manager
Holds credentials
Keeps secrets out of code and the repository.
Part 6
CloudWatch
Logs, metrics and alarms
You find out about a problem before the customer does.
These are common reference sketches, not a design for your system. The right choice depends on traffic, budget, team skills and compliance needs, and it is worth a conversation before anything is built.
Simulation: reference sketches only, not a quote or a design for your project.
Have a project in mind?
Turn what you just learned into a plan: answer a few questions and get a technical approach, risks and a roadmap for your own idea.