AI-Native Infrastructure · Agentic Runtime · Developer Advocate · CNCF Ambassador
Open-Source Ecosystem VP @ Dynamia

I work on how AI systems run in production.

My focus is not models, but the infrastructure, runtime, and engineering discipline required to make LLM and agent systems reliable, observable, and evolvable over time.

• AI-Native Infrastructure
  Runtime abstractions, resource scheduling, isolation, cost and lifecycle management for AI workloads.

• GPU Scheduling & Virtualization (HAMi)
  Building HAMi — GPU sharing/virtualization and scheduling for Kubernetes to improve utilization, governance, and multi-tenant efficiency.
  Project: https://github.com/Project-HAMi/HAMi

• Agentic & RAG Systems
  Retrieval-augmented architectures, stateful agents, evaluation and observability of long-running AI systems.

• Engineering Methodology
  Turning AI from demos into systems: clear boundaries, verifiable pipelines, and repeatable delivery patterns.

• Open Source & Ecosystems
  Developer platforms, community-driven infrastructure, and sustainable open-source governance.

• 12+ years in cloud-native and platform engineering
• CNCF Ambassador, long-term contributor in the Kubernetes / service mesh ecosystem
• Former Developer Advocate at Tetrate
• Founder of Cloud Native Community (China)

• Languages: Go, Python, TypeScript
• Platforms: Kubernetes-centric infrastructure, distributed systems
• AI Systems: RAG pipelines, agent runtime design, system observability

• Website: https://jimmysong.io
• GitHub: https://github.com/rootsongjc
• LinkedIn: https://linkedin.com/in/jimmysongio
• X: https://x.com/jimmysongio