The rapid advancement of large-scale language and vision models has significantly propelled the AI domain. We now see AI enriching everyday life in numerous ways – from community and shared virtual reality experiences to autonomous vehicles, healthcare innovations, and accessibility technologies, among others. Central to these developments is the real-time implementation of high-quality deep learning models, facilitated by the proliferation of mobile and embedded computing devices. These advancements have seamlessly integrated machine intelligence into various aspects of our lives. However, the efficient training and inference of Deep Neural Networks, especially in large transformer-based models, remain a major challenge due to high computational demands. This talk will address innovative, efficient deep learning techniques to overcome these hurdles. Zhenglun Kong’s focus on optimizing model performance in a sustainable manner, aligning with the goal of making AI accessible to all. By developing models that are not only powerful but also resource-efficient, we pave the way for AI’s integration into more diverse and resource-constrained environments, democratizing access to these transformative technologies