Dongdong Chen is a Principal Research Manager at Microsoft GenAI, leading multimodal foundation model development. Currently, he is leading the phi-3-vision model post-training and safety training . He delivered many great techniques into Microsoft products, including but not limited to designing CSwin transformer to help win the 26M 3P contract, leading and delivering Florence BackgroundRemoval and MagicEraser model into MS paint&Offices&Windows Photo&Windows Copilot, developing high-quality person segmentation for Teams Spatial Presenter and ImageText retrieval for Bing Ads. He has published more than 80 papers in top conferences and journals. His research interests focus on large-scale unimodal and multimodal pretraining, AIGC generative models, and AI security. He is the SPC of AAAI 2022, Area Chair of CVPR 2022, ICPR 2022, ECCV 2022, CVPR 2023, NeurIPs 2023 and the Associate Editor of Pattern Recognition.