Chao Wang
CTO of Tengyun Zhisuang
Chao Wang is currently the CTO of Tengyun Zhisuang, a GPU rental startup, and a technology leader with 15 years of experience in ICT product development and solutions. He has held technical and product positions at globally renowned companies including Tencent, Huawei, and Ericsson. During his time at Tencent, he led as the primary project owner for a mobile internet security product that reached 1 billion monthly active users and received multiple SVP President’s Awards. He is now focused on the field of AI infrastructure, where his team’s technical innovations at the inference framework level have been incorporated and cited by multiple open-source projects, with model downloads on Hugging Face surpassing 20K. He is dedicated to providing high-performance GPU computing services for global large-model application scenarios and driving the industrial application of AI technologies. Mr. Wang holds a Master of Science degree from the Hong Kong University of Science and Technology, as well as dual Bachelor’s degrees in Engineering and Literature from Xiamen University. He has also studied abroad at the University of Manchester in the UK and Pohang University of Science and Technology in South Korea, equipping him with a global perspective and cross-cultural technical collaboration capabilities.
Topic
End-to-End Intelligent Computing Power Services for Next-Generation Large Model Developers
Starting from the pain points faced by developers such as difficulty in accessing computing power, low utilization rates, and high usage costs, this presentation introduces the solution architecture of Tengyun ACC. It covers the full stack—from underlying supply chains and high-power data center infrastructure to cloud platform architecture and acceleration frameworks—to address the challenges developers face in using computing power for large model development. At the same time, in response to the rapidly growing demand for inference scenarios, Tengyun provides developers with inference frameworks such as ty-vllm and speculative inference services, helping improve inference efficiency and reduce costs in production environments. Outline: 1. Overview of Tengyun Zhisuang 2. Current pain points of large model developers and Tengyun’s solutions 3. Technical architecture of the Tengyun Zhisuang cloud platform 4. Core technical advantages of the Tengyun Zhisuang ACC solution 5. Team and case studies
