Background: Construction sites are complex ecosystems characterized by dynamic workflows, heterogeneous machinery, and diverse stakeholders. Traditional site management often suffers from inefficiencies, cost overruns, and sustainability challenges due to fragmented decision-making and limited use of real-time data. Emerging advances in Artificial Intelligence (AI), predictive analytics, and digital optimization provide new opportunities to transform construction site operations into integrated, intelligent and sustainable ecosystems.

Objectives: This thesis proposes and evaluates an AI-Driven Optimization Framework that addresses critical challenges in construction site ecosystems. The framework is designed to enhance operational efficiency through intelligent scheduling, resource allocation, and passage guidance; reduce downtime and costs by enabling predictive maintenance of multi-brand machinery; improve sustainability by minimizing fuel consumption, carbon emissions and material waste; and ensure scalability and interoperability across diverse construction contexts.

Methods: A Combined Formal Experiment and Case Methodology was adopted, combining reinforcement learning, predictive modeling, and simulation-based evaluation. Real-world construction datasets, IoT sensor inputs, and synthetic scenarios were utilized to train and validate the framework. Performance was assessed using established metrics such as task completion time, resource utilization, maintenance efficiency, and environmental impact.

Results: The framework achieved significant improvements across key dimensions. Predictive maintenance extended equipment life and reduced unplanned failures. AI-based passage guidance minimized travel distances, leading to lower emissions and fuel consumption. Intelligent scheduling enhanced coordination among workers and equipment, thereby improving productivity. Pilot validations confirmed the framework’s adaptability to diverse construction settings while demonstrating measurable cost savings and sustainability gains compared to conventional approaches.

Conclusions: The study confirms the feasibility and effectiveness of integrating AI-driven optimization in construction site ecosystems. By unifying predictive, prescriptive, and sustainable decision-making techniques, the framework not only enhances operational efficiency but also contributes to safer and more environmentally responsible construction practices. This research advances the digital transformation of the construction industry and lays the foundation for broader industrial adoption of AI-powered optimization frameworks.