Flori-Core Enterprise OS

Problem

High-altitude floriculture operations run physical field sensors, labour, spray logs, pack house, procurement, and export documentation as disconnected systems with no real-time visibility.

Approach

Built a multi-tenant 'Farm Operating System' connecting IoT field sensors to logistics and market operations through a unified backend, with real-time WebSocket gateways and an AI assistant.

Tech highlights

NestJS 11 + Next.js 16 monorepo (Turborepo) with Prisma across 22 migrations covering telemetry, labour, spray logs, pack house, procurement, export docs, and CRM; MQTT v5 via EMQX ingests cold-room sensor data into TimescaleDB for time-series compliance reporting; Socket.io gateway streams live operational events; AI assistant (Claude + OpenAI) via Vercel AI SDK; M-Pesa B2C for payroll disbursements; Redis, AWS S3, Prometheus + Sentry observability.

Results

Live demo with one-click sign-in across eight roles — managing director, HR manager, financial controller, head of production, QC lead, stores manager, export sales lead and lead driver — over sixteen modules spanning farm zones, production, operations, pack house, cold room, team, HR and training, logistics, inventory, stores, procurement, sales and CRM, compliance and financials, backed by 22 Prisma migrations.

Challenges

Ingesting high-frequency MQTT cold-room sensor data into TimescaleDB without dropping readings during connectivity gaps at remote high-altitude field sites — buffering and backfilling on reconnect — while keeping the Socket.io live dashboard responsive and the 22-migration schema coherent across telemetry, labour, spray logs, and export docs.