All work

Dashboards

Plant Monitor

Live plant-floor telemetry over a TimescaleDB hypertable, with continuous aggregates so a twelve-week view renders as fast as a one-hour view.

Role
Architecture, data model, full-stack build
Timeline
TBD
Status
Live
Client
Food & beverage manufacturer — plant operations
Plant Monitor interface
Client work shown anonymized — walkthrough on request

The problem

Line supervisors were reading machine conditions off local HMIs and transcribing them into spreadsheets. Nobody could compare this shift to last month, and by the time a trend was visible the product was already out the door.

The approach

Sensor readings land in a TimescaleDB hypertable with continuous aggregates rolled up at the intervals the floor actually asks for. The app reads the aggregate that matches the selected range instead of scanning raw rows, so range switching stays instant. Thresholds and alert states are configuration, not code.

The result

One live view of the floor that supervisors keep open on the wall display, with history that goes back far enough to argue with. Exact uptime and scrap deltas: TBD.

The stack

LovableTanStack StartTypeScriptTailwindshadcn/uiSupabaseTimescaleDBContinuous aggregates

Inside the app

Real screens with branding blurred, or reproduced with anonymized data — no client names, logos, or user identities.

Machine drilldown: one sensor against its threshold band, with every breach listed and timed.

Want one of these for your operation?

Tell me the outcome you need. I'll tell you what it takes, what it costs, and whether I'm the right person for it.