Northvane Energy
A live fault map for 40,000 sensors
Service
Web platform
Duration
20 weeks
Year
2025
Stack
React · Rust · Kafka · TimescaleDB
−62%
alert noise
40k
sensors streaming live
In one paragraph
An operations dashboard for a regional grid operator. Faults appear on the map within two seconds, and related alerts are grouped into one incident.
(04) — Case notes
The problem, what we built, and what changed. Numbers are from the client’s own analytics.
The problem
Northvane’s control room watched six tools at once. One storm could raise 3,000 alerts, most of them echoes of the same fault.
What we built
A streaming pipeline in Rust that groups related alerts into a single incident, and a map that shows every sensor, fault and crew in one view. It runs on the control room’s wall and on a crew lead’s phone.
The result
Alert noise fell 62%. The median time from fault to dispatched crew dropped from 19 minutes to 7.