LoginTel SaaS

The Challenge

LoginTel approached us with a critical problem: pharmaceutical and food distributors were losing millions due to temperature excursions during transport. They needed a robust IoT platform that could monitor cold chain conditions in real-time, trigger instant alerts, and generate compliance reports for regulatory bodies.

The key challenges were:

• Real-time reliability — IoT devices must stream temperature, humidity, and location data 24/7 without gaps
• Protocol complexity — Devices communicate via MQTT, TCP/IP, and proprietary binary formats
• Alert latency — Temperature violations must trigger SMS/email alerts within seconds, not minutes
• Compliance reporting — Detailed audit trails required for FDA, HACCP, and ISO certifications
• Multi-tenant architecture — Each logistics company needs isolated data, dashboards, and alert rules

Our Approach

We designed a three-tier architecture optimizing for both real-time data ingestion and user-facing performance:

IoT Data Pipeline (Python FastAPI)

• TCP parser service — Custom binary protocol decoder for legacy IoT devices
• MQTT broker integration — Sub-second ingestion from modern sensor nodes
• Time-series database — PostgreSQL with TimescaleDB extension for efficient historical queries
• Alert engine — Rule-based threshold monitoring with Prometheus AlertManager

Web Platform (Next.js + Strapi)

• Multi-tenant SaaS — Organization-scoped data isolation with role-based access control
• Real-time dashboard — WebSocket connections for live sensor updates
• Content management — Strapi headless CMS for device catalog, documentation, and compliance templates
• Report generation — Automated PDF exports with time-stamped temperature graphs and chain-of-custody logs

Observability Stack

• Prometheus — Metrics collection from IoT gateway, FastAPI workers, and Next.js server
• Grafana — Operational dashboards tracking ingestion rates, alert response times, and system health
• PagerDuty integration — Critical alerts escalate to on-call engineers automatically

The Stack

• Frontend: Next.js 14 with App Router, TanStack Query, Recharts for data visualization
• Backend API: Python FastAPI with async workers, Celery for background jobs
• CMS: Strapi v4 with PostgreSQL, custom content types for device profiles
• IoT Layer: MQTT broker (Mosquitto), custom TCP server for binary protocols
• Database: PostgreSQL with TimescaleDB for time-series data
• Monitoring: Prometheus, Grafana, AlertManager
• Infrastructure: Docker Compose for staging, Kubernetes for production, DigitalOcean managed PostgreSQL

Outcome

The platform launched in Q2 2024 and has since monitored over 2 million shipments across Morocco, France, and Spain:

• 99.9% uptime — Zero critical incidents in 18 months of production
• Sub-5-second alerts — Average alert delivery time of 3.2 seconds from threshold violation to SMS
• 100% audit compliance — All clients passed FDA and HACCP audits using LoginTel-generated reports
• 40% cost reduction — Clients reduced spoilage-related losses by eliminating undetected temperature excursions

Engineering Lessons

Protocol abstraction pays off: By decoupling device protocols (MQTT, TCP) from the business logic layer, we onboarded three new device models in under a week each.

Time-series matters at scale: TimescaleDB’s hypertable partitioning kept query response times under 200ms even with 6 months of sensor data per device.

Alert fatigue is real: We added smart throttling rules (e.g., “Only alert once per hour for non-critical thresholds”) after clients complained about notification overload during the first month.

Ready to build mission-critical IoT systems? Book a working session with CodeLovers.