A digital-twin solution that replaces Palantir Foundry for a water utility — SAP, GIS and SCADA/IoT data fused into a DataStax Graph operational twin, scored by watsonx.ai, and acted on through a touch-first, geospatial control plane. Start below.
Live demoThe operator's live map: tap assets, trace downstream impact, draw geospatial queries, approve human-validated repair orders. Dark / Satellite / Hybrid basemaps.
The full blueprint — digital-twin center, ontology, 12-subsystem catalog, detailed data-flow schematic, and the control-plane spec. Links to the live demo.
The detailed runtime, protocol & deployment wiring of every tier — plus the application-server (BFF) that connects the control plane to the back end, and the request/data-flow sequences.
Where it runs and how it survives — HA, on-prem, hybrid & SaaS topologies, per-tier RPO/RTO, and a hurricane-grade disaster-recovery plan (off-island DR + island-isolation mode).
The validated 6-pipeline data-integration spine (StreamSets) that feeds the twin — SCADA / SAP / GIS simulators → fusion → twin sink → predictive maintenance → repair router.
The leadership KPI dashboard — Resumen General regional service rollup (1.33M clients), predicted failures, generator readiness, and the BCR investment portfolio. Also embedded inside the control plane.
The same leadership content in the study mock-up's structure — Purpose/Evidence, criticality situation, financial & BCR, asset health, and the Ops Assistant — re-skinned in the project design.
Every Approve mints a work order automatically — track them all here with a live stage pipeline, per-stage detail, and immutable final states for audit & traceability.
An advanced, faceted explorer over the 39-node digital twin — filter by type, zone & risk band with live counts, search & sort, and open any asset's 360° profile (score breakdown, work orders).
The island's water loop in one view — 19 AAA reservoirs with alert tiers, production → distribution → consumption → sewer → treatment → recycled back to storage, plus rain capture & sanitization, with the friction (produced vs consumed vs lost, overflow, rejected) highlighted live.
The running proof of the two-store twin: a real DSE Graph (traversals + geohash spatial
index) with per-node pointers into IBM DB2 master records, hydrated on click by a Python BFF.
Deploys as one podman pod (stack-up-kube.sh, a Kubernetes-format manifest —
all components on localhost) or on a bridge via stack-up.sh.
The same 39-node twin rebuilt on a single PostgreSQL 16 + PostGIS database — recursive-CTE blast radius, exact GiST spatial queries (no prefilter/refine), master record in the same row (the two-hop collapsed), and the composed downstream ∩ region in one SQL statement. Run it side-by-side with the Graph × DB2 stack and compare query-for-query.