Automated Graph Machine Learning Graph neural networks (GNNs) have emerged as the dominant approach for graph machine learning. However, most GNNs are manually designed, requiring extensive human effort and lacking adaptivity. Over the past few years, there has been a significant surge of interest in automated graph machine learning, which aims to combine the power of AutoML and GNNs. This field has demonstrated impressive capabilities in enhancing the adaptability and generalizability of graph machine learning techniques. In this presentation, I will provide a concise overview of automated graph machine learning. Additionally, I will delve into the details of our recent research endeavors in graph neural architecture search, which currently stands as the most influential and promising area within automated graph machine learning. Specifically, I will discuss various aspects of graph neural architecture search, such as modeling diverse graph structures, scalability to large graphs, generalization under distribution shifts, and robustness to adversarial attacks. Moreover, I will introduce AutoGL, the first open-source library developed exclusively for automated graph machine learning.