Loading...
Publication
Refactor do sistema de reward tokens
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 4.99 MB | Adobe PDF |
Advisor(s)
Abstract(s)
O presente projeto foi desenvolvido na empresa Blip, no âmbito do Mestrado em Engenharia Informática do ISEP, e responde a limitações identificadas no sistema de reward tokens da plataforma FanDuel. Este sistema, utilizado para personalizar e distribuir recompensas a milhões de utilizadores, apresentava elevados níveis de duplicação de código, dificuldade de manutenção e complexidade na criação de novos modelos de rewards. O principal objetivo foi o desenvolvimento de uma arquitetura modular e padronizada, baseada em microsserviços, que facilitasse a reutilização de componentes e a introdução eficiente de novos tokens. Para alcançar este objetivo, a abordagem combinou práticas ágeis com a metodologia Design Science Research (DSR), utilizando tecnologias como Java, Spring Framework, Docker, Kafka e Couchbase. A solução foi validada por meio de testes unitários, de integração, end-to-end, bem como uma avaliação estatística com a equipa. Os resultados demonstram uma redução de cerca de 75% na duplicação de código, promovendo uma base mais modular e reutilizável. O tempo médio de criação de novos modelos de reward passou de 4 a 8 semanas (dependendo da complexidade do modelo e dos tokens associados) para 3 a 6 semanas, representando uma melhoria significativa na eficiência do desenvolvimento. Além disso, verificou-se uma redução estimada de 40% nos erros de integração, contribuindo para uma maior escalabilidade, manutenção e fiabilidade do sistema, segundo a avaliação técnica da equipa de engenharia da Blip. Como contributo principal, a solução posiciona a Blip para responder de forma mais eficaz às
exigências do mercado, reforçando a sua capacidade de inovação e alinhamento com as melhores práticas de engenharia de software e com os objetivos estratégicos da Flutter. Este trabalho estabelece, também, uma base sólida para futuras evoluções técnicas e organizacionais do sistema de rewards. No entanto, a solução poderá exigir adaptações adicionais em contextos com requisitos mais voláteis ou com sistemas legados complexos, o
que se sugere com o trabalho futuro.
This project was developed at Blip as part of the Master’s in Informatics Engineering at ISEP, in response to limitations identified in the reward token system used by the FanDuel platform. This system, critical for the personalized distribution of rewards to millions of users, suffered from significant code duplication, maintenance challenges, and complexity in the creation of new reward models. The main objective was to develop a modular and standardized architecture, based on microservices, which would facilitate the reuse of components and the efficient introduction of new tokens. To achieve this, the approach combined agile practices with the Design Science Research (DSR) methodology, using technologies such as Java, Spring Framework, Docker, Kafka and Couchbase. The solution was validated through unit, integration and end-to-end tests, as well as a statistical evaluation with the team. The results show a reduction of around 75% in code duplication, promoting a more modular and reusable base. The average time taken to create new reward models went from 4 to 8 weeks (depending on the complexity of the model and the associated tokens) to 3 to 6 weeks, representing a significant improvement in development efficiency. In addition, there was an estimated 40% reduction in integration errors, contributing to greater scalability, maintenance and reliability of the system, according to the technical assessment of Blip's engineering team. As its main contribution, the solution positions Blip to respond more effectively to market demands, reinforcing its capacity for innovation and alignment with the best software engineering practices and Flutter's strategic objectives. This work also lays a solid foundation for future technical and organizational evolutions of the rewards system. However, the solution may require additional adaptations in contexts with more volatile requirements or complex legacy systems, which is suggested as future work.
This project was developed at Blip as part of the Master’s in Informatics Engineering at ISEP, in response to limitations identified in the reward token system used by the FanDuel platform. This system, critical for the personalized distribution of rewards to millions of users, suffered from significant code duplication, maintenance challenges, and complexity in the creation of new reward models. The main objective was to develop a modular and standardized architecture, based on microservices, which would facilitate the reuse of components and the efficient introduction of new tokens. To achieve this, the approach combined agile practices with the Design Science Research (DSR) methodology, using technologies such as Java, Spring Framework, Docker, Kafka and Couchbase. The solution was validated through unit, integration and end-to-end tests, as well as a statistical evaluation with the team. The results show a reduction of around 75% in code duplication, promoting a more modular and reusable base. The average time taken to create new reward models went from 4 to 8 weeks (depending on the complexity of the model and the associated tokens) to 3 to 6 weeks, representing a significant improvement in development efficiency. In addition, there was an estimated 40% reduction in integration errors, contributing to greater scalability, maintenance and reliability of the system, according to the technical assessment of Blip's engineering team. As its main contribution, the solution positions Blip to respond more effectively to market demands, reinforcing its capacity for innovation and alignment with the best software engineering practices and Flutter's strategic objectives. This work also lays a solid foundation for future technical and organizational evolutions of the rewards system. However, the solution may require additional adaptations in contexts with more volatile requirements or complex legacy systems, which is suggested as future work.
Description
Keywords
Keywords (Topic): Microservices Reward Systems Personalized Tokens Modular Architecture Scalability Online Betting User Retention Keywords (Technologies): Java Spring Framework Docker Couchbase RESTful API Kafka Microserviços Sistemas de rewards Tokens personalizados Arquitetura modular Escalabilidade Apostas online Retenção de utilizadores