On November 24th, the "Embodied Intelligence Industry-Academia-Research Integration Ecosystem Conference," hosted by Qingdao City College, was successfully held in Qingdao. The conference brought together technical experts from leading industry enterprises to discuss technological breakthroughs, talent cultivation systems, and industry implementation pathways in the era of embodied intelligence.

Shanghai Zentek Intelligent Technology Co., Ltd. (hereinafter referred to as "Shanghai Zentek"), as one of the invited companies, officially signed an agreement with Qingdao City College to jointly establish the "Qingdao Embodied Intelligence Industry-Academia-Research Base." Additionally, company technical personnel delivered a keynote report at the main forum, sharing the latest advancements in AI infrastructure and agent applications.

Signing of the Embodied Intelligence Industry-Academia-Research Base

In a highlight of the conference, the Qingdao Base for Embodied Intelligence Industry-Academia-Research Integration was officially signed and established! Shanghai Zentek, as one of the first seven embodied intelligence ecosystem enterprises, completed the on-site signing alongside industry representatives such as Unitree Technology, DOBOT, AgileX Robotics, Lingcifang Robot, and AIO Intelligent.

In the future, Shanghai Zentek will engage in in-depth collaboration with Qingdao City College around embodied intelligence engineering. The establishment of this base marks a shift in the embodied intelligence industry from "enterprise-led" to "deep industry-academia-research collaboration," injecting new momentum into regional and national industry development.

Zentek Technical Experts Deliver Presentation

At the conference, technical presentations were delivered successively by representatives from NVIDIA, Unitree Technology, Ubtech, Zhiyuan, Yinhe Tongyong, and others. Invited to the stage, Zentek technical personnel, based on the CGC platform, shared insights into GPU computing power and application practices in the era of intelligent agents, covering functional dimensions such as the NIM microservice architecture, intelligent agents, and Blueprint workflows.

CGC Platform: A One-Stop Gateway for AI Microservices

Lichan CGC, as a unified platform in China for downloading NVIDIA NIM (NVIDIA Inference Microservice), provides a complete pipeline encompassing model downloads, microservice deployment, and application workflows. It can serve as the infrastructure for agent development in both academic and industrial settings. Over a hundred NVIDIA NIMs are already available on CGC, covering areas such as large language models, text-to-image, video, medical, engineering, and manufacturing.

CGC Direct Link: cgc.chancloud.com

Technical Presentation: Core Capabilities of the CGC Platform

01

NIM: Standardized Inference Microservice Architecture
The technical team highlighted the NIM resource library within CGC, where various models provide services through unified APIs and environments, facilitating rapid invocation and migration for teaching and R&D. 

LIM (Lichan Inference Microservice) offers model contribution and management workflows, supporting the deployment of self-developed or fine-tuned models on CGC in a standardized format, thereby enriching the local model ecosystem.

02

 Spark Zone: Tailored Support for DGX Spark
Considering the Arm architecture characteristics of Spark, CGC provides specifically adapted NIM images and deployment methods, enabling stable operation of large models and agent pipelines even on desktop-level inference devices.


03

AI App Market: One-Click Intelligent Agent Workflows
Scenario-based applications composed of Blueprints like RAG, VSS, and digital humans provide ready-to-use intelligent agent examples suitable for classroom teaching, scientific research validation, and enterprise prototype development.



Shanghai Zentek Exhibition Booth Debut






At the conference venue, Shanghai Zentek Intelligent Technology set up a dedicated exhibition booth, showcasing the highly anticipated NVIDIA DGX Spark desktop AI inference device. As an efficient desktop computing platform for developers, the actual Spark unit attracted a large number of students, faculty, and industry guests upon its display.






NVIDIA DGX Spark is a desktop AI inference device for developers. Based on the new GB10 architecture and equipped with 128GB unified memory, it can support local inference for large models up to the 200B scale. With its compact size, low noise, and no need for complex deployment, it provides an efficient, ready-to-use model inference environment for university labs, research teams, and individual developers.

On-site visitors engaged in in-depth exchanges with the Zentek technical team. Many educators showed strong interest in Spark's advantages for lightweight deployment in teaching and research, while students actively experienced its inference speed and development workflow. The exhibition booth maintained a lively atmosphere with consistently high visitor interaction.



Technology Empowers Education, Education Drives Industry

As embodied intelligence technology increasingly permeates real-world industrial scenarios, engineering practical ability has become a new core focus in talent cultivation. The success of this conference marks a solid step forward in building the embodied intelligence education system and demonstrates a tangible path for industry-academia-research collaboration to more universities, enterprises, and students.





In the future, Shanghai Zentek will continue to collaborate with universities and ecosystem partners to jointly promote the growth of China's embodied intelligence talent ecosystem, truly converging technological innovation and talent development into a force driving the industry forward!


DGX Spark Direct Purchase Link: https://item.jd.com/10156720695244.html