Working on open source frameworks, such as Tensorflow and Pytorch, to create an AI model is feasible, however, managing the end-to-end life cycles of AI data/models for enterprise environment is a different story. In this talk, I will discuss about how we built our technologies to manage data and models using our Brightics AI platform, which provides ways to collect, prepare, augment, and label data and makes it easy to train AI models on multiple servers, share them with your peers, and deploy them on edge devices. Technical components of our AI platform, for example, auto-labeling training data, creating better confidence metric for deep learning outputs, and proposing best pre-trained models for a given data set - will be explained. Finally, I will discuss enterprise use-cases of deep learning and reinforcement learning and what we learned from building an AI platform at Samsung SDS.

Seungjai Min is a Master at Samsung SDS where he leads AI Research Center. He was a post-doc at Lawrence Berkeley National Laboratory and an research engineer at LG Electronics Research Center. He received his Ph.D. degree in Electrical and Computer Engineering from Purdue University, his master's degree in Electonics Engineering from Seoul National University and B.S. degree in Electonics Engineering from Sogang University.Seungjai Min is a Master at Samsung SDS where he leads AI Research Center. He was a post-doc at Lawrence Berkeley National Laboratory and an research engineer at LG Electronics Research Center. He received his Ph.D. degree in Electrical and Computer Engineering from Purdue University, his master's degree in Electonics Engineering from Seoul National University and B.S. degree in Electonics Engineering from Sogang University.