Arquitetura escalável para sistemas de delivery: uma abordagem baseada em eventos, cache distribuído e indexação geoespacial

Abstract

This work presents the development of a backend platform for a food delivery marketplace system, designed to efficiently manage the entire order lifecycle, from product selection to delivery and service evaluation. The proposed solution adopts a modular monolith architecture with strong adherence to Hexagonal Architecture principles, promoting a clear separation between business logic and external dependencies. Furthermore, the system incorporates concepts of Event-Driven Architecture, enabling asynchronous processing of critical tasks through distributed messaging, which enhances scalability and resilience. To support performance, in-memory caching technologies and state management strategies are employed, reducing latency in critical operations. Additionally, the system leverages geospatial indexing based on H3 to optimize the delivery dispatch process, enabling efficient proximity queries. The adopted methodology is based on the definition of functional and non-functional requirements, domain-driven design modeling, and incremental implementation using containerization to ensure environment consistency. The results indicate that the solution is capable of handling high-demand scenarios, providing adequate performance, architectural organization, and ease of evolution. It is concluded that the combination of clean architecture, asynchronous processing, and geolocation techniques constitutes an effective approach for building modern delivery systems, serving as a solid foundation for both real-world applications and academic studies.