APCS (Algerian Port Community System) was inspired by the inefficiencies we observed in traditional port operations—manual booking processes, capacity bottlenecks, and time-consuming gate entries that slow down logistics chains. We built this full-stack solution using React for an intuitive role-based frontend, Laravel as a robust API backbone with MySQL for data persistence, and a Python FastAPI microservice powered by Llama 3.2 for intelligent assistance. The system integrates QR code-based gate access, real-time capacity monitoring, and smart late-arrival handling that automatically reschedules bookings when trucks miss their slots. Throughout development, we faced challenges with Docker orchestration across three services, implementing secure inter-service communication between the AI agent and backend using token-based authentication, and optimizing the LLM's response time while maintaining context awareness through our custom tool system. We learned valuable lessons about microservice architecture, the power of agentic AI workflows, and how to balance real-time system requirements with user experience. The result is a production-ready terminal management system that reduces booking approval time , eliminates manual gate processing, and provides instant operational insights through natural language queries.
Log in or sign up for Devpost to join the conversation.