| Authors: |
Evgeny Polyachenko, Jorge Augusto Meira, Renato Silva de Melo, Daniel Antunes, Nikolai Riumin, Gabriel Irribarem Soares Ruas, Carlos Andre Zavadinack, Alexander Fowler, Kristof Horvat and Radu State |
| Abstract: |
The Carrier Communication Hub (CCH) is a production-ready system that applies Industry 4.0 principles to solve the challenge of carrier communication for a major European digital logistics platform. The system addresses the operational bottleneck of manual carrier management by providing a unified, automated framework to handle interactions with a network of over 750 carrier companies. It integrates machine learning, data-driven recommendation systems, and automated bidding processes to optimize shipping operations across Europe.
Built as a standalone Symfony 7.1+ application, the CCH architecture is designed for robustness and interoperability. It features deep, bidirectional integration with an ODOO CRM front-end, utilizes a central database for order and carrier data, and leverages a multi-channel communication infrastructure (Email, WhatsApp). The system’s asynchronous processing is handled by Symfony Messenger with Redis/SQS message queues to ensure resilient performance under high communication loads. Operationally, CCH offers two distinct modes: a direct Order Mode, where agents use an ML-based pricing engine for estimates, and a competitive Bidding Mode. The bidding is a two-round auction, progressing from an open to a targeted round with a system-suggested price to ensure market-competitive rates.
The core of CCH's intelligence lies in its carrier selection and price prediction mechanisms. The recommendation system uses multi-stage filtering (geography, vehicle type) followed by a weighted scoring algorithm that evaluates carriers on dimensions including Recent Activity Level, Price Competitiveness, and Route Variety. The price prediction engine utilizes a portfolio of machine learning models tailored for different vehicle classes and accommodates special customer configurations with fixed-rate formulas or model feature flags.
The system is production-deployed, with a containerized architecture and a comprehensive REST API to facilitate integration and support operational scaling. The CCH project thus demonstrates a functional application of a modular software architecture, dual operational modes, and machine learning to create a robust framework for automating and optimizing large-scale carrier management. |