teaching

Winter 2025/26: Introduction to Computational Sustainability (191.021)

TU Wien · elective course in master programmes

This course covers two perspectives: Sustainable AI (reducing the energy/CO₂ impact of AI systems) and AI for Sustainability (using AI to address climate and resource challenges).

Course topics

• Sustainable AI: energy & CO₂ of training/inference; measurement/profiling; LLM efficiency; reducing cost via model compression, scheduling, and efficient serving.
• Edge–cloud ML: inference and learning across the edge–cloud continuum; latency and data quality constraints in distributed settings.
• AI for sustainability: smart ICT & data-center optimization; renewable/grid-energy management; climate-impact applications (e.g., water quality, flood sensing).

Students develop skills in profiling and interpreting energy/latency metrics; reasoning about accuracy–latency–energy trade-offs in large-scale and edge–cloud ML systems; and applying ML methods to sustainability-driven use cases.

Role: course organization; 1–2 lectures; assignment design & grading; examination.