He Zhongjiang

Chairman and General Manager at China Telecom Artificial Intelligence Technology Co Ltd

Speaker

He Zhongjiang, graduated from Tsinghua University with a master's degree in Business Administration. He is currently serving as the Chairman and General Manager of China Telecom Artificial Intelligence Technology Co., Ltd., and is a dual-appointed scientist at the Shanghai Artificial Intelligence Laboratory.

He has been engaged in big data and AI work for a long time, having held positions such as the Director of the Data Management Department of the Enterprise Informatization Business Division (Data Center) of China Telecom, Deputy General Manager of the Cloud Network Operations Department (Big Data and AI Center) of China Telecom, and Deputy General Manager of the Digital Intelligence Technology Branch Company of China Telecom.

His main work achievements and scientific research accomplishments include: establishing the China Telecom AI research and development team; leading the team to build a four-level AI computing power system for China Telecom, promoting the establishment of a digital intelligence branch company and an artificial intelligence company, and creating the first domestic digital customer service representative; developing the Xinghe AI empowerment platform to support the development of China Telecom's entire network big data AI business; developing semantic, speech, image, and multi-modality Xingchen series large models, achieving independent control of key technologies of large models. He has developed semantic large models with hundreds of billions and thousands of billions of parameters, where the hundred-billion-parameter model ranked seventh overall and first in the hundred-billion-parameter model category on the well-known C-Eval list, and the hundred-billion-parameter model won first place in the text generation track of the well-known OpenCompass list; developed a hundred-billion-parameter multi-modality large model, launched 15 text-to-image and 55 image-to-image services, achieving top-five results on the well-known MME and MMBENCH lists; developed a hundred-billion-parameter visual large model, achieving object detection driven by natural language, supporting a billion-scale feature base and fine-grained analysis, deploying core algorithms through the Xinghe platform nationwide, empowering more than 100 city governance downstream tasks, with the algorithm being called 400 million times per day; mastered end-to-end full-stack digital human core technology, supporting the national application of China Telecom's digital human customer service, serving more than a million users; led the team to win 22 championships and runner-ups in international AI top competitions, and published more than ten papers in international AI top conferences.