The EvoClass team has won the Society and Sustainability track championship at the Huawei Developer Competition 2025 Asia Pacific final. Its award-winning project, EvoClass, applies multi-agent AI to teaching and learning, connecting research with an education platform already used in a university course.
A track championship among 192 regional entries
According to Huawei Cloud Hong Kong's competition report, the final pitch took place on 19 November 2025 at The Hong Kong Polytechnic University, followed by the awards ceremony on 20 November. The competition attracted nearly 1,000 developers and 192 projects from 10 countries and regions.
Fifteen teams advanced to the final: nine in Society and Sustainability and six in Finance. Finalists presented their projects and answered questions from judges with investment, academic and technology backgrounds. The EvoClass team received the Society and Sustainability championship, a category recognizing applications with social value.
The research team behind EvoClass
The EvoClass team is based in PolyU's Department of Data Science and Artificial Intelligence. Prof. Ran Cheng leads the project, with PhD students Beichen Huang, Zhenyu Liang and Bowen Zheng as core members. Their work brings AI-agent research into an educational setting.
Bringing multi-agent AI into learning
EvoClass is designed around a learning environment in which AI agents develop alongside students. PolyU describes a platform that provides feedback informed by learning context and automates repetitive teaching tasks, with the aim of making personalized support more accessible.
The educational question behind this approach is practical: how can learners receive help that fits what they are studying, while teachers retain time for explanation, discussion and thoughtful course design? This gives the technology a clear purpose within teaching and learning.
Classroom practice at PolyU
PolyU reports that EvoClass has been deployed in DSAI2201: Data Structures and Algorithms. The university's announcement describes improved student engagement and reduced educator workload in that setting, without providing numerical results.
This classroom experience connects the project to real teaching needs. A course also provides a setting for examining whether support is useful across different topics, how students respond to feedback and where educators need more control. These are important questions for evaluating educational AI beyond a competition demonstration.
From competition recognition to educational value
The championship is a milestone for the EvoClass team. Its longer-term significance lies in the educational goals behind the project: useful learning feedback, more room for teachers to focus on teaching, and broader access to support. For EvoClass, those goals give context to the award and a basis for discussing the platform's development.
Further reading
Read PolyU's announcement of the EvoClass team's award for the team and classroom background.
See Huawei Cloud Hong Kong's competition report for the final's schedule, track structure and results.