OASIS HIVE Limited
Home Projects Pricing Contact

Utilities ยท Rift Valley Energy

Metering data that finally reached the people who needed it

A smart metering platform ingesting 2.4 million readings a day, with theft detection and outage alerts built on top.

Client

Rift Valley Energy

Sector

Utilities

Duration

Six months

Delivered

2025

2.4M
Readings per day
-18%
Technical losses
11 min
Average outage detection
+34%
Billing accuracy

The problem

Meters collected readings but the data sat in regional silos. Billing ran on estimates, losses were high and unexplained, and outages were discovered when customers called to complain.

What we did

  • Built a streaming ingestion layer handling 2.4 million readings per day with backfill and replay capability.
  • Consolidated regional silos into a single modelled warehouse with consistent definitions for consumption and loss.
  • Trained a consumption anomaly model that flags likely meter tampering and bypass for field investigation.
  • Added near-real-time outage detection by correlating meter connectivity with grid topology.
  • Gave field teams a mobile view so investigations start with evidence rather than a hunch.

The outcome

2.4M

Readings per day

-18%

Technical losses

11 min

Average outage detection

+34%

Billing accuracy

“We stopped guessing at where the losses were coming from. The field teams now arrive already knowing what they are looking for.”

— Head of Operations, Rift Valley Energy

Technology used

Kafka Apache Spark ClickHouse Python PyTorch Grafana

Next step

Let us talk about your version of this problem

Every one of these started as a conversation about something that was not working properly.