Embracing AI for Sustainable Computing

The digital revolution is transforming scientific disciplines across the board, with our growing reliance on large-scale, heterogeneous, and often uncertain datasets propelling scientific discovery to new realms. This surge in data-intensive computing has underscored the critical need for sustainable, efficient, and intelligent data analytics systems. The IT industry is estimated to contribute 20% to the world's energy consumption by 2030, with an economic cost of over $100 Billion and unprecedented carbon emissions. This motivated all major IT companies to set goals for reducing their energy consumption and carbon footprints in the next decade. In this talk, I will present my research efforts in developing a seamlessly distributed and sustainable cyberinfrastructure enabled by AI for efficient search, access, and discovery of a vast amount of distributed, heterogeneous, complex, and dynamic data sources. I will illustrate how my work interweaves cutting-edge systems research, data science, and AI/ML technologies to develop resilient, intelligent, and sustainable solutions for large-scale data analytics pipelines. Our unique approach drastically reduces the energy consumption and carbon footprint of computing without negatively impacting the system's performance.