IoT / hardware
Sensor networks, edge computing, device control loops.
Sensor → gateway → cloud
Classic IoT pipeline. Constrained devices ship data via a local gateway to a cloud backend that handles storage and analytics.
doc { cols: 4 }
icon :sens tabler/temperature "Sensor"
icon :gw tabler/router "Gateway"
icon :cld tabler/cloud "Cloud"
icon :dash tabler/chart-dots "Dashboard"
sens --> gw "Bluetooth"
gw --> cld "MQTT"
cld --> dashRobotic control loop
Sense → plan → act, repeated every tick. The planning step is where an ML model might sit; the rest is deterministic.
doc { cols: 4 }
icon :world tabler/leaf "Environment"
icon :sense tabler/camera "Sense"
icon :plan tabler/brain "Plan"
icon :act tabler/engine "Actuate"
world --> sense
sense --> plan
plan --> act
act --> world "effect"Smart home topology
Hub-and-spoke over the local network. The hub bridges to the cloud for off-site control and voice assistants.
doc { cols: 3, rows: 3 }
icon :hub @B2 tabler/home-bolt "Hub" sizeScale=1.3
icon :lights @A1 tabler/bulb "Lights"
icon :lock @C1 tabler/lock "Lock"
icon :cam @A3 tabler/camera "Camera"
icon :therm @C3 tabler/temperature "Thermo"
hub <-> lights
hub <-> lock
hub <-> cam
hub <-> thermEdge + cloud inference
Latency-critical inference runs at the edge; aggregate analytics go to the cloud. Models retrained in the cloud and pushed back to the edge on a schedule.
doc { cols: 4, rows: 2 }
icon :sens @A2 tabler/camera "Camera"
icon :edge @B2 tabler/device-tv "Edge"
icon :cloud @D2 tabler/cloud "Cloud"
icon :model @D1 tabler/brain "Retrain"
sens --> edge "stream"
edge --> cloud "metrics"
cloud --> model "new data"
model --> edge "deploy"