About

About This Site

This is an original technical site maintained by Jimmy Song, focusing on systematic research and engineering practice in AI-Native Infrastructure, GPU Virtualization, and Heterogeneous Computing Governance.

The core question this site addresses is not model parameters or short-term technology trends, but rather how AI workloads can achieve stable, controllable, and efficient computing power usage in real production environments through proper infrastructure abstraction and resource governance mechanisms. Related topics include but are not limited to: heterogeneous computing resource models (especially GPU utilization efficiency), scheduling and isolation mechanisms, platform engineering, system governance, and the deployment of open source projects in industrial environments.

All content is organized based on long-term engineering practice, research writing, and public discussion. There is no content aggregation and no chasing of short-term trends.

About the Author

Jimmy Song is an architecture researcher and open source ecosystem builder who has long been deeply engaged in cloud native and AI-native infrastructure.

His story really starts back in university. He graduated from Wuhan University of Technology with a degree in software engineering. He was already a hardcore Linux user back then, not just for writing code but as the actual operating system on his daily computer. Ubuntu was his favorite at the time.

His first stop after graduation was iFlytek, where he worked as a software engineer building the speech cloud big data platform. The stack was mostly Java and Shell, and Linux was, of course, a daily companion.

After that he joined TalkingData as an operations engineer, still working on big data platforms, this time with Hadoop, Spark, Docker, and Kubernetes. It was also during this period that he picked up Go.

What really changed the direction of his career was learning Kubernetes. Through Kubernetes he found his way into its community, got to know the CNCF, and in 2018 became a CNCF Ambassador. From then on, open source wasn’t just part of his work, it became part of his identity, and he started doing things like Developer Advocate, work that connects technology with people.

Looking back, the thread is clear: from CNCF, to cloud native, to today’s AI-native infrastructure. He has been moving along this line the whole way.

Currently, he serves as Open Source Ecosystem VP at Dynamia, responsible for ecosystem building, technical narrative, and community collaboration around the CNCF project HAMi, focusing on GPU virtualization, heterogeneous computing scheduling, and AI scenario resource governance.

He is also a CNCF Ambassador and the founder of the Cloud Native Community (China).

Interests & Life

Outside of work, he is someone who likes to keep busy and explore.

He is a drone enthusiast. He has been flying since 2018 and has logged over a thousand kilometers in total. Seeing the world from a different angle is something he has always found addictive.

He also loves road trips and travel, especially to islands and forests. Hitting the road toward quiet places with great scenery is his favorite way to unwind.

Two other things make him genuinely happy: cooking, especially making and eating fish, and hanging out with his two cats, who are a big part of his life.

Research & Practice Focus

His long-term focus centers on the following directions:

• AI-Native Infrastructure: Re-examining infrastructure design assumptions at the resource modeling, scheduling, and governance levels from the non-deterministic nature of AI workloads
• Computing Governance & GPU Virtualization: Improving resource utilization efficiency in heterogeneous hardware environments, researching GPU sharing, isolation, scheduling, and governance mechanisms
• Cloud Native Architecture Evolution: Focusing on Kubernetes’s capability boundaries under AI workloads, and the evolution direction of platform engineering and system governance
• Open Source Ecosystem & Standardization: Promoting engineering-verifiable, design-evolvable open source collaboration, and facilitating deep connections between Chinese and global communities

These directions point to a core question: How to evolve computing power from “scarce hardware resources” into governable, schedulable, and scalable infrastructure capabilities.

Writing & Publishing

He has authored or participated in translating multiple cloud native and architecture-related books, including “Deep Dive into Istio” and “Future Architecture”. In recent years, his main writing focus has shifted to AI-Native Infrastructure, GPU Virtualization, and Computing Governance, and he continues to organize and publish in the form of books, columns, and long-form articles.

Relationship Between This Site and His Work

The content on this site represents his personal engineering and architectural thinking in AI-Native Infrastructure and Computing Governance. These insights form the long-term methodological foundation for his work at Dynamia.ai.

For company-related products, business, and official commitments, please refer to information released through Dynamia’s official channels.

Collaboration & Contact

At this stage, he mainly works on GPU virtualization, heterogeneous computing scheduling, and open source ecosystem collaboration around Dynamia and its related open source projects.

He only maintains long-term, constructive communication and collaboration with the following types of topics:

• Technical and ecosystem discussions related to AI-Native Infrastructure, computing governance, and GPU virtualization
• Community collaboration, engineering practice, and standardization exchange at CNCF / open source project levels
• Research-oriented discussions on medium-to-long-term technical evolution or high-quality public sharing

If necessary for communication, please leave a message through this site’s Contact page. He will respond based on the direction and maturity of the topic.