Tencent’s Weixin Integrates Ray for Large-Scale AI Deployment

Lawrence Jengar
Jul 02, 2025 13:55

Tencent’s Weixin team has embraced Ray and Kubernetes to enhance their AI infrastructure, tackling challenges in resource utilization and deployment complexity.

Tencent’s Weixin team has taken significant strides in their AI infrastructure by deploying Ray, an open-source distributed computing engine, alongside Kubernetes. This integration aims to address the challenges of deploying large-scale AI systems efficiently and cost-effectively, according to Anyscale.

Ray’s Role in AI Infrastructure

The Weixin team, responsible for the popular Chinese app serving mainland users, has faced numerous technical hurdles, including resource utilization, deployment complexity, and application orchestration. The team sought a solution that could handle their extensive AI computing needs, which span content recommendation, product operations, and content creation.

Ray, developed by UC Berkeley’s RISELab, has gained traction as a leading distributed computing framework. It simplifies the development of distributed applications with its intuitive programming model, allowing the Weixin team to efficiently manage large-scale AI workloads.

Challenges and Solutions

Weixin’s existing infrastructure faced limitations in handling computationally intensive tasks, such as Optical Character Recognition (OCR), which require over a million CPU cores. The P6n platform, while suitable for responsive online tasks, proved costly and complex for large-scale deployments. On the other hand, the Gemini platform, optimized for offline processing, fell short in meeting real-time performance needs.

To overcome these challenges, Weixin developed AstraRay, a new AI compute engine built on Ray. AstraRay addresses cost efficiency, high throughput, and reduced deployment complexity, enabling scalable AI deployment across heterogeneous resources.

Ray’s Integration and Impact

Ray’s integration into Weixin’s infrastructure has enabled the development of AstraRay, which supports ultra-large-scale resource scheduling and efficient deployment. AstraRay boasts enhancements over the community version of KubeRay, including support for millions of nodes and improved resource utilization.

By leveraging Ray’s capabilities, Weixin has streamlined its AI operations, reducing the complexity of deploying AI applications and enhancing performance. This integration not only optimizes resource use but also prepares Weixin for future AI advancements.

Future Prospects

With the successful deployment of AstraRay, Tencent’s Weixin is well-positioned to expand its AI capabilities. The project, initiated a year ago, continues to evolve, setting the stage for more sophisticated AI applications and innovations in the coming years.

Image source: Shutterstock