This repository contains the full system design, specifications, architecture, and implementation for an intelligent SCADA solution targeted at small water systems (typically < 1,000 connections). The system combines edge computing (Raspberry Pi), hybrid AI models, and a clean multi-tenant web interface to deliver reliable monitoring with minimal data transfer costs.

Project Goals

Provide affordable, professional-grade monitoring for small water systems that cannot justify traditional SCADA costs.
Minimize ongoing data egress costs through intelligent local processing and tiered telemetry.
Enable both non-technical board members and licensed operators to have appropriate visibility and control.
Build a foundation for proactive alerting, anomaly detection, and long-term system health insights using AI.

Key Concepts

Monitored Objects: First-class abstraction for pumps, filters, tanks, sensors, etc.
Tiered Telemetry: Immutable core signals + optional signals (full telemetry vs alert-only mode).
Hybrid AI Architecture: Small edge models for local decisions + larger cloud models for complex reasoning.
Low-Egress Design: Heavy summarization locally, rich context only sent during alerts.
Multi-Tenancy: Supports both individual system owners and professional operators managing multiple systems.

Repository Structure

All project documentation lives under /docs:

docs/
├── vision.md
├── architecture/
├── specs/
├── hardware/
├── decisions/          # Architecture Decision Records (ADRs)
└── images/

Active development tasks are tracked in /tasks.

Agent role definitions and swarm coordination live in /agents and follow the Hierarchical Agent (HA) methodology defined in this template.

Current Status

This project is in the early architecture and specification phase. The core vision, use cases, and high-level system design are being finalized before moving into detailed implementation.

Maintained by Ty_Tech