Core platform · MDMS

Grid Data Management Platform From AMI reads to certified bills — and everything the meter data touches.

A complete Meter Data Management System: streaming ingestion, a three-phase VEE engine with weather-aware and machine-learned estimation, an exception workbench, master data from your systems of record, and certified, immutable data served to billing, outage operations, gas, device commands, market settlement, analytics, and regulators.

Kafka-native

Streaming Ingestion Pipeline

3-phase VEE

Validate · Estimate · Act

Electric + Gas

One pipeline, Commodity-aware

Billing-grade

Certified · Immutable · Auditable

The core of the family

The system of record between the meter and the bill

Everything a utility decides — what to bill, how to treat reads through an outage, which device to disconnect, what to settle with the market — is only as good as its meter data. GDMP exists to make that data trustworthy: ingested without loss, validated by explicit rules, estimated transparently, worked through a real exception workbench, and served consistently to every consumer — then certified into bills that never change without a revision trail.

Streaming ingestion

Interval reads, register reads, and signals flow in over governed Kafka topics into versioned day facts — idempotent, durable, with DLQ replay governance and full support triage tooling.

VEE engine

Validation rule families for register, interval, cross-channel, and extreme-value checks; action rules that estimate, flag, open exceptions, notify, or raise service orders — all behind an eligibility gate that explains every decision.

Weather-aware & ML estimation

Estimation by reference day, load-profile regression, temperature correlation, and interval-register conversion — with a Spark ML estimation service and MLflow pipelines for temperature-segmented models.

Exception management lake

A workbench for the exceptions VEE raises: assignment pools and rules, escalation policies, manual override and suppression, bulk resolution, trends, a command-day graph, and a map.

Master data sync (SDR)

Continuous synchronization of customers, contracts, premises, devices, and channels from the systems of record — staged, validated, conformed with SCD history, and traceable payload by payload.

Projection & decision serving

Raw facts, signals, and read decisions become current serving rows with replay and convergence checks — VEE, Billing, Outage, Gas, dashboards, and APIs all see one consistent answer.

Billing module

Determinants for energy, demand, TOU, reactive, ratchet, CPP/RTP, and special contracts; certification and cycle close; CIS export with profiles and acknowledgements; daily extract; rebill, dispute, and hold; NEM credits; immutable ledgers; shadow reconciliation for cutover.

Outage module

A first-class outage record with lifecycle, scope, evidence, and policy; outage-aware VEE treatment; holds and automatic restoration replay; outage-linked exception resolution; PSPS handling; and a command center with map.

Gas commodity

Pressure districts, governed conversion factors (JSON, CSV, EDI 867), HDD and diurnal-profile derivation, gas VEE kernels, safety events, unaccounted-for gas, gas billing certification, dashboards, and ten standard reports.

Coordinated Control Operations

Device commands — disconnect, reconnect, on-demand reads, pings — with CCO assignments, device criticality, request statistics, and a clean billing pull-request boundary.

Virtual & unmetered meters

Formula-driven virtual meters, submeter and check-meter context, and unmetered supply — the network the utility actually has.

Search & master-data workspace

Elasticsearch-backed global, master-data, and exception search; analyst inquiry over customers, premises, devices, and attributes with comments, tags, and personalization — with Whisper-Wind embedded for plain-English questions.

Operations Monitor

A standalone, read-only monitoring application: pipeline health, queue lag, worker metrics, runtime-stage logs, Grafana dashboards, and an alert catalog with drill-down queries.

Market settlement

A companion settlement service: zonal load and forecasts, capacity tag sets and peak events, LMP, UFE and loss factors, supplier tranches, approvals and market submissions — from the same certified usage the bills came from.

Rate analysis

What a customer’s actual interval usage would cost under alternative rate schedules — analyzed from the platform’s own serving data.

Admin, security & deployment

Feature catalog, role bindings, navigation, bill cycles, day types, gate policies, retention, and system configuration through Keycloak with schema-scoped isolation; a client configuration workbook, bootstrap package, readiness gates, and rollback runbook for every new deployment.

Field operations & Green Button

Manual reads, meter tests, service actions, and regulated data sharing with privacy aggregation.

Admin & access control

A feature catalog with role bindings, feature groups, and per-role navigation — governed through Keycloak.

How it works

See it, end to end

Every read is validated, every estimate is auditable, and every consumer sees the same converged answer.Head-endsystemsinterval readsregister readsmeter eventsvia KafkaIngestionVEEvalidate · estimate · actProjectionServingDLQ replay · exception lake · outage-aware treatmentMaster data spine (SDR sync)service points · meters · devices · channels — synchronized from systems of recordBilling & CIS exportOutage · CCO · gasSettlement · APIsEvery read is validated, every estimate is auditable, and every consumer sees the same converged answer.
Ingest

Step 1

Ingest

Head-end systems publish reads and signals onto governed Kafka topics; workers land every payload idempotently, with traceability and DLQ replay.

Assemble

Step 2

Assemble

Day facts are versioned and correlated with the master-data picture — customer, premise, device, channel.

Validate & act

Step 3

Validate & act

The VEE engine runs its three phases under an eligibility gate; findings become estimates, flags, exceptions, notifications, or service orders, with outage policy applied to affected days.

Project & serve

Step 4

Project & serve

Decision and serving rows refresh with replay and convergence checks — every downstream consumer reads one consistent, current answer.

Certify & deliver

Step 5

Certify & deliver

Billing determinants are certified and exported to CIS, daily extracts stream, settlement and analytics draw from the same serving data — every step observable in the Operations Monitor.

Use cases

Where Grid Data Management Platform earns its keep

AMI rollout backbone

Collect, validate, and store data from millions of smart meters as the authoritative meter data repository.

Billing-grade data supply

Deliver certified determinants to CIS with every estimate, edit, rebill, and hold auditable — and prove parity against the legacy system before cutover.

Outage-aware operations

Treat meter data correctly through outages and PSPS events, replay on restoration, and close the exceptions an outage caused in bulk.

Multi-commodity operations

Run electricity and gas on one platform, with commodity-specific derivation, VEE, safety, and billing where it matters.

Remote device operations

Coordinate disconnects, reconnects, and on-demand reads through the same governed data spine.

Retail and wholesale settlement

Settle supplier load against market data from the same certified usage the bills came from.

Regulatory compliance

Green Button sharing, privacy aggregation, and full audit trails for regulator-ready reporting.

Revenue assurance

Spot missing reads, stuck meters, and estimation drift before they become revenue leakage.

Under the hood

Technology

Python / FastAPI

Kafka

PostgreSQL + TimescaleDB

Elasticsearch

Angular 20

Spark ML + MLflow

Grafana

Keycloak

Governed worker runtime

Docker / Kubernetes

Better together

Part of the Grid Data Family

Every family application shares one identity platform, one design system, one registration pattern, and one deployment path.

See Grid Data Management Platform live

A working demonstration takes thirty minutes — on our environment or yours, with realistic simulated data.

Connect With Us