Computing toward 2030: algorithmic contributions to the sustainable development goals

This thesis presents algorithmic solutions for sustainable development across three interconnected domains, organized as three pillars: sustainable mobility and drone logistics, the environmental impact of geo-distributed data centers, and intersectional approaches to women's safety in urban spaces. Pillar I --- Sustainable Mobility and Drone Logistics. We address the scheduling of parcel deliveries for a UAV fleet operating over a decentralized network of service stations providing battery replacement. This could be considered a more sustainable alternative to traditional truck deliveries. We formalize MiKe (Minimum maKespan), prove its NP-hardness, and propose P&D (Partition & Deliver), a polynomial-time heuristic that outperforms the state of the art by up to 45% in delivery completion time. We then extend the framework to the k-Home k-Drone Delivery problem and propose RedMat, the first approximation algorithm with a provable optimality bound for drone delivery makespan minimization, and its dynamic variant RedMatDyn, which handles online demand arrivals and variable wind conditions, improving mean delivery time by up to 24.5% over the state of the art. Pillar II --- Environmental Impact of Geo-Distributed Data Centers. Data centers have become a foundational component of modern ICT infrastructure, driven by the widespread adoption of online services and cloud computing paradigms such as IaaS, PaaS, and SaaS. Yet this growth raises growing concerns about their environmental impact. We argue that cloud sustainability cannot be reduced to carbon emissions alone. We define formal sustainability profiles for data centers encompassing carbon, water, and land use footprints, grounded in an analysis of real provider sustainability reports. We propose an optimization-based cloud orchestrator for multi-factor job placement and migration, and conduct an extensive simulation study of spatial and temporal workload shifting using real workload traces and grid mix data. Results show reductions of up to 85% in carbon, 50% in water, and 45% in land use footprint. Pillar III --- Intersectional Approaches to Women's Safety in Urban Spaces. We critically analyze safe path planning algorithms through an intersectional lens, revealing systematic failures including reliance on biased crime data, gender-binary assumptions, and neglect of accessibility. We propose design principles for equitable safety technologies that center diverse identities, community knowledge, and structural change over purely technical optimization.