Code · Internet · Electronics
Gear we recommend
The boards and parts we reach for, and what to look for when you buy them. We update it as we build.
Some links may be affiliate links: if you buy through them we may earn a small commission at no extra cost to you. It never changes what we recommend.
Microcontrollers and boards
Where most IoT builds start.
ESP32 DevKit (ESP32-WROOM-32)
Wi-Fi and Bluetooth on a cheap board; our default for connected sensors and relays.
ESP32-S3 DevKit
More memory and USB built in; good for cameras, displays and small on-device AI.
Arduino Uno R4 WiFi
The easiest way to learn; huge community and shields that just work.
Raspberry Pi 5
A full Linux computer: gateways, cameras, dashboards and local services.
Raspberry Pi Pico W
Tiny and cheap with Wi-Fi; great for MicroPython projects.
Sensor starter kit
Temperature/humidity (DHT22, BME280), motion (PIR), relays and jumper wires to prototype fast.
Local AI workstation parts
For running open models on your own machine. VRAM decides what fits; everything else keeps it fed and cool.
GPU, 12 GB VRAM
Entry point: small and quantised 7–8B models run comfortably.
GPU, 16 GB VRAM
The sweet spot for many teams: 13–14B models quantised, longer context.
GPU, 24 GB VRAM or more
Larger quantised models and room for several users at once.
64 GB system RAM
Headroom for loading models, embeddings and everything else running.
2 TB NVMe SSD
Models are large; fast storage keeps load times short.
850 W+ 80 Plus Gold PSU and good airflow
GPUs draw power in bursts and run hot; this is where cheap builds fail.
Not sure what to buy?
A short hardware spec review saves expensive mistakes.