Pular para o conteúdo

O conteúdo principal desta página está em inglês.Traduzir esta página

IA e temas em alta

IoT (Internet of Things) roadmap

IoT engineers build connected devices: smart homes, farm sensors, factory monitors and wearables. Learn microcontrollers, networking protocols, cloud platforms, security and edge AI.

Tempo
⏱ 4–6 months
Level
Beginner
Steps
6 steps
Architecture ↓Cheat sheet ↓4 interview Q&As ↓
Compartilhar

The roadmap

Click a step to open the full guide.

Best architecture: an IoT platform

From devices to dashboards, with security at every hop.

IoT reference architecture

  1. Devices (sensors, edge AI)
  2. Gateway / connectivity
  3. MQTT broker (TLS, device certs)
  4. Rules & stream processing
  5. Time-series DB
  6. Dashboards · alerts · OTA updates

Why it's built this way

  • Unique certificates per device and TLS everywhere.
  • Edge processing sends events, not raw streams.
  • OTA updates with signing and staged rollout.

IoT (Internet of Things) cheat sheet

The essentials on one screen. Bookmark it for revision.

Protocols

Wi-FiHigh bandwidth, higher power
BLELow power, short range (phones, wearables)
LoRaWANVery long range, tiny data, low power
MQTTPub/sub messaging over TCP
Matter / ThreadSmart-home interoperability standards

Sensor buses

I2C2 wires, many devices, addresses
SPIFast, more wires
UARTSimple serial, 2 devices
Analog (ADC)Read voltages, e.g. soil moisture

Improve performance

  • Batch readings and send less often to save power and bandwidth.
  • Process at the edge; send events, not raw streams.
  • Use QoS levels wisely; QoS 2 is slowest.
  • Downsample old time-series data for long-term storage.

Security checklist

  • No default passwords; unique credentials per device.
  • TLS for MQTT/HTTP; verify server certificates.
  • Signed firmware with secure boot and rollback protection.
  • Separate IoT network; close unused ports and services.
  • Plan for updates over the device's whole life.

Most important interview questions

Click a question to see a strong answer. Each step also has its own questions.

1O que é o IoT?

Networks of physical devices with sensors and connectivity that collect data and act on the world, usually with cloud or edge processing.

2Describe a typical IoT architecture.

Devices and sensors → gateway/connectivity (Wi-Fi, BLE, LoRa) → message broker (MQTT) → cloud processing and storage → dashboards, alerts and apps.

3How do you handle devices that go offline?

Buffer data locally, retry with backoff, use last-will messages in MQTT to detect disconnects, and design the cloud to tolerate gaps.

4Which wireless technology would you pick for a farm sensor 2 km away?

LoRaWAN: long range, very low power, small payloads, ideal for periodic readings.

Related roteiros

6

Robotics

6 steps · 6–9 months · Intermediate

Build robots that sense, think and move.

View roadmap →
6

Cloud Engineer

6 steps · 5–7 months · Intermediate

Design, build and run systems on AWS, Azure or Google Cloud.

View roadmap →
6

AWS

6 steps · 6–8 weeks · Intermediate

The most widely used cloud platform.

View roadmap →
1

IoT (Internet of Things) · Step 1 of 6 · 3 weeks

Microcontrollers and sensors

Read the physical world.

The ESP32 is the most popular IoT board: cheap, with Wi-Fi and Bluetooth built in. Learn how sensors connect (I2C and SPI buses) and how to read them reliably.

MicroPython is quicker to learn; C++ (Arduino framework) is more common in production firmware.

What you'll learn

  • ESP32 and Arduino
  • Digital and analog inputs
  • I2C, SPI and UART
  • Common sensors (temperature, humidity, motion)
  • Relays and actuators
  • MicroPython vs C++

A smart sensor node

  1. Sensor
  2. ESP32 reads
  3. Process
  4. Send over Wi-Fi
  5. Cloud / dashboard

Key concepts

Microcontroller
A tiny computer on one chip that runs one program directly on hardware.
I2C / SPI / UART
Common ways sensors talk to microcontrollers.
GPIO
General-purpose pins for digital input/output.

Step by step

  1. 1.1

    Set up ESP32

    Install the toolchain (Arduino IDE / PlatformIO or MicroPython).

  2. 1.2

    Read a sensor

    Temperature/humidity over I2C; print every few seconds.

  3. 1.3

    Control an output

    Switch an LED or relay based on the reading.

Notes

  • Check sensor voltage (3.3 V vs 5 V) before connecting.

Tips & tricks

  • Start with a weather station: temperature, humidity and a display.

Interview questions

I2C vs SPI?

I2C uses 2 wires and supports many devices with addresses (slower). SPI uses more wires but is faster, good for displays and SD cards.

The quick way to study this

Read a temperature sensor and print values every few seconds.

2

IoT (Internet of Things) · Step 2 of 6 · 3 weeks

Connectivity and protocols

Get data off the device.

MQTT is the standard IoT messaging protocol: devices publish small messages to topics on a broker, and apps subscribe. It is lightweight and works on poor networks.

For smart homes, Matter is the new standard that lets devices from different brands work together.

What you'll learn

  • Wi-Fi and Bluetooth Low Energy
  • MQTT and brokers
  • HTTP and WebSockets
  • LoRaWAN for long range
  • Matter and Thread for smart homes

MQTT publish/subscribe

MQTT broker
  • Temperature sensor (publish)
  • Door sensor (publish)
  • Dashboard (subscribe)
  • Database (subscribe)
  • Phone alerts (subscribe)

Key concepts

MQTT
Lightweight publish/subscribe messaging for devices.
Broker
The server that routes MQTT messages between publishers and subscribers.
QoS
Delivery guarantee: 0 at most once, 1 at least once, 2 exactly once.

Step by step

  1. 2.1

    Run a broker

    Start an MQTT broker locally or use a managed one.

  2. 2.2

    Publish

    Device publishes JSON readings to home/livingroom/temp.

  3. 2.3

    Subscribe

    An app subscribes and reacts or stores data.

Notes

  • Use QoS levels in MQTT to control delivery guarantees.

Tips & tricks

  • Run a local MQTT broker on your laptop to practise.

Interview questions

Why MQTT instead of HTTP for IoT?

MQTT keeps a lightweight persistent connection, has tiny message overhead, supports publish/subscribe and works well on unreliable, low-bandwidth networks.

The quick way to study this

Publish sensor readings over MQTT and view them in a dashboard.

Matter and Thread explained →
3

IoT (Internet of Things) · Step 3 of 6 · 3 weeks

Cloud and dashboards

Store, visualise and act on data.

IoT data is time-series: readings over time. Store it in a time-series database and show it in dashboards with alerts.

Over-the-air (OTA) updates let you fix firmware on deployed devices without visiting them.

What you'll learn

  • IoT cloud platforms
  • Time-series databases
  • Dashboards (Grafana and others)
  • Alerts and rules
  • Device management and OTA updates

IoT data pipeline

  1. Devices
  2. MQTT broker
  3. Ingest rules
  4. Time-series DB
  5. Dashboard + alerts

Key concepts

Time-series data
Values recorded over time, like readings every minute.
OTA update
Updating device firmware over the air.
Device shadow / twin
A cloud copy of a device's last known state.

Step by step

  1. 3.1

    Store

    Write readings to a time-series database.

  2. 3.2

    Visualise

    Build a dashboard with live charts.

  3. 3.3

    Alert

    Rules that notify you when values cross thresholds.

  4. 3.4

    OTA updates

    Update firmware remotely and safely.

Notes

  • Plan OTA updates from the start; adding them later is hard.

Tips & tricks

  • Set an alert that notifies your phone when a value crosses a threshold.

Interview questions

How do you update firmware on thousands of devices safely?

Signed images, staged rollout (small group first), A/B partitions with automatic rollback if the new firmware fails health checks, and monitoring.

The quick way to study this

Build a dashboard with a phone alert for your weather station.

AWS roadmap →
4

IoT (Internet of Things) · Step 4 of 6 · 3 weeks

Power and hardware design

Make devices last and survive.

Battery devices must sleep most of the time and wake briefly to measure and send. A power budget tells you how long batteries will last.

Moving from breadboard to a custom PCB makes devices smaller and more reliable.

What you'll learn

  • Deep sleep and power budgets
  • Batteries and solar
  • PCB design basics
  • Enclosures
  • Reliability and watchdogs

Battery-friendly device cycle

  1. Wake up
  2. Read sensors
  3. Send batch
  4. Deep sleep

Repeats as a loop

Key concepts

Deep sleep
Almost everything switched off to save battery.
Duty cycle
The share of time the device is awake.

Step by step

  1. 4.1

    Power budget

    Measure current in active and sleep modes.

  2. 4.2

    Deep sleep

    Wake, measure, send, sleep.

  3. 4.3

    Hardware

    Move from breadboard to PCB and enclosure.

Notes

  • Wi-Fi uses a lot of power; send data in batches.

Tips & tricks

  • Measure real current draw with a meter, not just datasheets.

Interview questions

How do you make a battery IoT device last months?

Keep it in deep sleep most of the time, batch transmissions, use low-power radios (BLE, LoRa), lower sampling rates, and measure real current draw.

The quick way to study this

Make your sensor run for a month on batteries using deep sleep.

5

IoT (Internet of Things) · Step 5 of 6 · 2 weeks

Segurança

Protect devices and data.

Insecure IoT devices are a favourite target for attackers. Never ship default passwords, always encrypt traffic and sign firmware updates.

Put IoT devices on a separate network from computers and phones.

What you'll learn

  • Unique credentials per device
  • TLS encryption
  • Secure boot and signed firmware
  • Network segmentation
  • Updating devices

IoT security layers

  1. Secure boot & signed firmware
  2. Unique device identity
  3. Encrypted communication (TLS)
  4. Least-privilege cloud access
  5. Network segmentation & monitoring

Key concepts

Secure boot
Device only runs firmware signed by you.
Network segmentation
Keeping IoT devices on their own network.

Step by step

  1. 5.1

    Unique credentials

    Per-device certificates or keys; no default passwords.

  2. 5.2

    Encrypt

    TLS for all traffic.

  3. 5.3

    Sign firmware

    Secure boot and signed updates.

  4. 5.4

    Isolate

    Separate network for IoT devices.

Notes

  • Default passwords are the most common IoT weakness.

Tips & tricks

  • Use per-device certificates for cloud connections.

Interview questions

What are the biggest IoT security risks?

Default or shared passwords, unencrypted traffic, unsigned firmware updates, exposed management ports and devices that are never patched.

The quick way to study this

Add TLS and unique credentials to your project.

6

IoT (Internet of Things) · Step 6 of 6 · 3 weeks

Edge AI

Run AI on the device itself.

Edge AI runs small models directly on devices: detecting a wake word, spotting machine faults from vibration, or counting people with a camera. It is faster, cheaper and more private than sending everything to the cloud.

Models must be small and quantised to fit microcontroller memory.

What you'll learn

  • TinyML
  • Model quantisation
  • Keyword spotting and anomaly detection
  • Camera-based detection
  • Edge vs cloud trade-offs

Edge AI vs cloud AI

Edge

  • Fast, works offline
  • Private
  • Small models only

Cloud

  • Big models
  • Needs connectivity
  • Data leaves device

Key concepts

TinyML
Machine learning that runs on microcontrollers.
Anomaly detection
Spotting readings that don't look normal.

Step by step

  1. 6.1

    Collect data

    Record real sensor data with labels (normal vs fault).

  2. 6.2

    Train small

    Train a tiny model; quantise it.

  3. 6.3

    Deploy on device

    Run inference on the microcontroller; send only events.

Notes

  • Collect real data from your device to train edge models.

Tips & tricks

  • Start with a vibration anomaly detector; it is simple and very practical.

Interview questions

When would you run AI on the device instead of the cloud?

When you need low latency, offline operation, privacy, or to save bandwidth and power by sending only events instead of raw data.

The quick way to study this

Train a tiny model to detect an unusual pattern in your sensor data.

Robotics roadmap →

Compartilhe esta página