Our research in causal machine learning spans causal discovery, causal representation learning, and causal inference and generation. In causal discovery, we develop methods to learn DAGs/Bayesian networks/SCMs from data, including both local and global methods, learning under insufficient data, and with heteroscedastic additive noise assumptions. For causal representation learning, we develop methods combining causal learning with deep learning to learn representations that are robust to data selection bias, latent confounders, and generalize well across domains. For causal inference and generation, we develop methods that equip generative models with causal reasoning capabilities, enabling faithful interventions and counterfactual generation under structural causal models, with applications in domain generalization, data augmentation, and out-of-distribution generalization.