My primary research interests lie at the intersection between distributed systems, machine learning and big data processing. Much of my work has focused on building scalable and self-tuning systems for large-scale machine learning and big data processing. Recently, I have been focusing on designing distributed systems for training big vision and language models, large graph neural networks and deep reinforcement learning models. This has led to publications at prominent conferences including OSDI, NSDI, USENIX ATC, VLDB and CoNEXT, as well as popular open-source machine learning software including KungFu, TensorLayer and HyperPose.