← ALL CASE STUDIESAI & DATA PLATFORMS · 2025

Grid telemetry at scale.

How a regional utility ingests two million smart meters a day — on a footprint 40% cheaper than the vendor quote.

CLIENTRegional utility
SERVICESData platforms, cloud, SRE
TIMELINE11 months, team of 6
HEADLINE RESULT40% below vendor quote
RJP-2025-009 · TELEMETRY INGESTALL SYSTEMS NOMINAL
2.0MMETERS INGESTED DAILY · 99.98% COMPLETENESS
INGEST ▸ OKSTREAM ▸ OKANOMALY ▸ OKARCHIVE ▸ OK

01The challenge

The utility's smart-meter rollout was outrunning its data platform. The incumbent vendor quoted a seven-figure annual bill for an ingestion tier that still dropped readings during storms — exactly when the grid team needed them most. Anomaly reports arrived a day late, as batch jobs.

02The approach

We replaced the batch pipeline with a streaming architecture: Kafka ingestion sized for storm-day peaks, time-series storage tiered by access pattern, and anomaly detection that scores readings in-stream instead of overnight. Every component is boring, proven, and sized from measured load — not the vendor's sizing guide.

BEFORE
NIGHTLY BATCH
DROPPED READS · DAY-LATE ALERTS
AFTER — STREAMING PLATFORM
INGEST
STREAM
ANOMALY
ARCHIVE

03The result

2.0MMETERS INGESTED DAILY
40%BELOW VENDOR QUOTE
<60sANOMALY DETECTION LATENCY

Grid operators now see anomalies in under a minute, storm-day ingestion holds at 99.98% completeness, and the platform runs on infrastructure the utility's own team can operate.

"We expected a data project. What we got was a team that argued us out of half the spend and delivered twice the platform."
GD
GRID OPERATIONS DIRECTOR, CLIENT NAME
NEXT CASE STUDYHospital zero-trust rollout
Start a project