Open-source projects for large-model data engineering, multimodal agents, real-time digital humans, and intelligent terminal applications.
DataScale-AI is an open-source organization initiated by Professor Jun Yu's team at the University of Science and Technology of China (USTC). It focuses on data engineering for large models, multimodal agents, real-time digital humans, and intelligent terminal applications, continuously developing open textbooks, algorithm frameworks, inference systems, and application projects.
The team is affiliated with the USTC-Huawei ICT Academy, the Multimodal Interaction Laboratory of the National Engineering Research Center of Speech and Language Information Processing, the Multimedia Computing and Intelligent Robotics Research Center, and the Joint Research Center for Multimodal Agents.
- Data Engineering for Large Models: an open textbook and project-based handbook for large-model data engineering.
- OpenTalking: an open-source real-time digital human framework.
- OmniRT: a multimodal generation inference framework for digital-human pipelines.
DataScale-AI is led by Professor Jun Yu, Associate Professor and Doctoral Supervisor at the Department of Automation, USTC, Distinguished Professor at Anhui Provincial Hospital, Huawei Most Valuable Instructor (MVI), and dual-certified Huawei/MindSpore developer evangelist. His long-term research focuses on multimedia computing and intelligent robotics.
Personal homepage: Prof. Jun Yu