The evolution of Generative Artificial Intelligence (GenAI) models is transforming the way software projects are developed with an impact on the whole development lifecycle, providing new tools to support analysis, development, and validation activities. However, using these technologies through a conversational approach does not guarantee reproducible results or an adequate level of control, essential in enterprise software development. This thesis, presents the study and application of a structured methodology for integrating the company's GenAI Plugin into a project related to the implementation of a cloud-based data platform for a client operating in the retail sector. The project involved migrating data from legacy systems to a Google Cloud Platform infrastructure to centralize business data, support business reporting, and develop a vertical application for the management of DESADV documents. The main contribution of this work is the definition of an engineering-oriented methodology for the use of GenAI. The approach is based on dedicated development workspaces, configuration and mapping files as the single source of truth, and operational guidelines that ensure reproducibility, traceability, and controlled software evolution. This enabled the GenAI Plugin to support activities such as requirements analysis, infrastructure development, SQL implementation, debugging, testing, and documentation. The results obtained demonstrate that, when properly governed, GenAI can evolve from a simple conversational assistant into a structured engineering tool. Although based on a naturally non-deterministic model, with a controlled engineering methodology, it is possible to obtain consistent, reproducible, and traceable tools and results, improving software quality, reducing development effort, and facilitating project maintenance. The experience gained suggests that this approach can be successfully adopted in other data engineering and enterprise software development projects involving GenAI technologies.
L'evoluzione dei modelli di Intelligenza Artificiale Generativa (GenAI) sta trasformando il modo in cui vengono sviluppati i progetti software con un impatto su tutto il loro lifecycle, fornendo nuovi strumenti a supporto delle attività di analisi, sviluppo e validazione. Tuttavia, l'utilizzo di tali tecnologie attraverso un approccio conversazionale non garantisce risultati riproducibili, né il livello di controllo richiesto nei contesti di sviluppo software in ambito enterprise. Questa tesi, presenta lo studio e l'applicazione di una metodologia strutturata per l'integrazione del plugin aziendale di GenAI in un progetto di sviluppo di una cloud data platform per un cliente nel settore retail. Il progetto ha previsto la migrazione dei dati da sistemi legacy verso un'infrastruttura basata su Google Cloud Platform, con l'obiettivo di centralizzare i dati aziendali, supportare la realizzazione di reportistica e sviluppare un'applicazione verticale per la gestione dei documenti DESADV. Il contributo consiste nella definizione di una metodologia ingegneristica nell'utilizzo della GenAI. L'approccio si basa sulla creazione di workspace dedicati, l'utilizzo di file di configurazione e mapping come unica fonte di verità e definizione di linee guida operative volte a garantire riproducibilità, tracciabilità ed evoluzione controllata del software. Questo ha consentito al plugin di GenAI di supportare attività come analisi dei requisiti, sviluppo dell'infrastruttura, scrittura delle query SQL, debugging e testing. I risultati ottenuti dimostrano che, se opportunamente governata, la GenAI può evolvere da semplice assistente conversazionale a strumento ingegneristico strutturato. Pur basandosi su un modello naturalmente non deterministico, con la metodologia proposta è possibile ottenere dei tool e risultati consistenti e tracciabili, migliorando la qualità del software, riducendo l'effort di sviluppo e semplificando le attività di manutenzione. L'esperienza maturata suggerisce infine che tale approccio possa essere adottato anche in altri progetti di data engineering che prevedano l'integrazione di tecnologie di GenAI.
Study and implementation of a generative AI-supported engineering approach for enterprise cloud data platform development
LOMBARDO, MATILDE
2025/2026
Abstract
The evolution of Generative Artificial Intelligence (GenAI) models is transforming the way software projects are developed with an impact on the whole development lifecycle, providing new tools to support analysis, development, and validation activities. However, using these technologies through a conversational approach does not guarantee reproducible results or an adequate level of control, essential in enterprise software development. This thesis, presents the study and application of a structured methodology for integrating the company's GenAI Plugin into a project related to the implementation of a cloud-based data platform for a client operating in the retail sector. The project involved migrating data from legacy systems to a Google Cloud Platform infrastructure to centralize business data, support business reporting, and develop a vertical application for the management of DESADV documents. The main contribution of this work is the definition of an engineering-oriented methodology for the use of GenAI. The approach is based on dedicated development workspaces, configuration and mapping files as the single source of truth, and operational guidelines that ensure reproducibility, traceability, and controlled software evolution. This enabled the GenAI Plugin to support activities such as requirements analysis, infrastructure development, SQL implementation, debugging, testing, and documentation. The results obtained demonstrate that, when properly governed, GenAI can evolve from a simple conversational assistant into a structured engineering tool. Although based on a naturally non-deterministic model, with a controlled engineering methodology, it is possible to obtain consistent, reproducible, and traceable tools and results, improving software quality, reducing development effort, and facilitating project maintenance. The experience gained suggests that this approach can be successfully adopted in other data engineering and enterprise software development projects involving GenAI technologies.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_07_Lombardo_Tesi.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
3.32 MB
Formato
Adobe PDF
|
3.32 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/261204