Learn with AI.
Verify before you trust it.

What Node actually is

JavaScript running outside a browser can read files and answer HTTP, and it has no DOM.

Modules and npm

Require a built-in module, and read package.json as the manifest of a package.

After What Node actually is

The request-response cycle

The plain Node HTTP server is the whole job: listen, read the request, write the response.

After Modules and npm

Why a framework

The same Hello response, first in raw Node and then with one Express route.

After The request-response cycle

Routing

Path parameters, query parameters, and a POST route, each as one real request.

After Why a framework

Middleware

The same two functions produce a different body when you swap which one is registered first.

After Routing

Request and response basics

express.json reads the body. res.status().json() sends one JSON response.

After Middleware

Documents and collections

Compare a SQL table with a MongoDB collection, then find one seeded document.

After Request and response basics

CRUD operations

Each operation runs against a fresh in-memory collection seeded inside the example.

After Documents and collections

Connecting Express to MongoDB

The request hits Express, the handler reads documents, and the response is the result.

After CRUD operations

Resource-oriented API design

Endpoints are nouns. GET, POST, PUT, PATCH, and DELETE are different promises, and PUT is idempotent in a way POST is not.

After Connecting Express to MongoDB

Headers, status codes, and pagination

Practice 201, 204, 404, and 409, then paginate, filter, and sort a collection with the query string.

After Resource-oriented API design

API versioning basics

A short introduction to keeping an old response shape working while a new one adds fields.

After Headers, status codes, and pagination

Structuring a real backend

Separate the route, the controller, the service, and the data function, even when they still live in one file.

After API versioning basics

Configuration and logging

Read configuration from one function, and write a log line as data instead of a sentence.

After Structuring a real backend

Embedding vs referencing

Compare embedding with referencing, then run one-to-one, one-to-many, and many-to-many shapes.

After Configuration and logging

Indexes and query performance

Compare a collection scan with an index scan on the same filter, then project away fields you do not need.

After Embedding vs referencing

Never trust client input

Reject a missing or non-string title before you write anything, and treat that check as the default.

After Indexes and query performance

Rate limiting and abuse basics

Count requests before the work, return 429 when the window is full, and notice why an AI route cares.

After Never trust client input

Environment and secrets

Read secrets from the environment, refuse to print them, and keep them out of the repository.

After Rate limiting and abuse basics

Reading a real stack trace

Read a thrown error from the throw site back through the callers, then return that message on the request.

After Environment and secrets

Structured logging

Compare a bare console.log with one JSON line that names the level, the note, the request, and the status.

After Reading a real stack trace

Calling an AI service from your backend

Make an HTTP request in the chat-completions shape and return the message content to the caller.

After Structured logging

Handling streamed responses

Read Server-Sent Events line by line, join the deltas, and ignore the done marker.

After Calling an AI service from your backend

Validating AI output before it reaches a user

Refuse a missing or blank summary, and do not forward the provider JSON just because it arrived.

After Handling streamed responses

Rate limiting and cost control for AI endpoints

Stop a request that would exceed a token budget, and do not call the model on that path.

After Validating AI output before it reaches a user