Introduction
Your phone unlocks with your face, your camera blurs backgrounds, your keyboard suggests the next word, and your smartwatch detects irregular heartbeats — often without sending anything to a server. That's edge ML: running models on the device itself ("at the edge" of the network) instead of in a data centre.
Why do it on the device?
| Benefit | Example |
|---|---|
| low latency | camera effects in real time; no network round trip |
| works offline | translation on a plane, factory sensors with no internet |
| privacy | voice and photos never leave the phone |
| lower cost | no server bill for billions of predictions |
| less bandwidth | a camera sends alerts, not hours of video |
The catch: devices have little memory, limited compute and batteries. A model that needs a data-centre GPU won't run on a smartwatch. So edge ML is mostly about making models smaller and faster without losing much accuracy.