Smart Water IoT Analytics Platform — Multi-Tenant Water Management

The problem with water management is simple. You dont know you have a leak until the monthly bill shows up, and by then you have lost 20-30% of your water supply. Utilities and property managers were flying blind, checking meters manually, reacting to complaints instead of data. I built Smart Water from scratch to fix that.

The platform ingests millions of meter readings per day through a third-party IoT provider API. Every reading hits DynamoDB, and DynamoDB Streams trigger Lambda functions for real-time alert evaluation. No polling, no batching delays. When a meter shows continuous flow at 3am, the system flags it immediately. I built 15+ alert types covering the scenarios that actually matter: CONTINUOUS_FLOW for leaks, BURST_SPIKE for pipe failures, NIGHT_USAGE for waste, REVERSE_FLOW for contamination risk, DAILY_VOLUME_ANOMALY using statistical analysis, plus device health alerts like METER_OFFLINE, LOW_BATTERY, and POOR_SIGNAL.

The architecture is multi-tenant from the ground up. Single-table DynamoDB design with composite keys and GSIs, supporting a Tenant to Unit to Meter hierarchy with proper data isolation. Each tenant can configure their own alert thresholds because a burst spike at a hotel looks very different from one at a residential complex. Cognito handles auth with role-based access for platform admins, tenant admins, and unit owners.

The dashboard is a Next.js 14 app with Recharts for data visualization, showing consumption trends, alert heatmaps, and device health in real time. Multi-language support for English, Arabic, and Bengali with full RTL layout. The whole thing runs serverless, so infrastructure costs scale with actual usage. Water providers using the platform catch leaks within minutes instead of months. That 20-30% water loss reduction is not a projection, its what happens when you give operators real-time visibility instead of monthly billing cycles.