HAMi-core and KAI Scheduler: GPU Sharing Moves from Allocatable to Governable
GPU sharing is moving from soft allocation to governance—scheduling decisions and runtime isolation can finally be reconciled.
Jimmy Song – Blog
Recent content in Blog on Jimmy Song
Latest posts
After HAMi Became a CNCF Incubating Project: Open Source Is Moving from Code to Consensus
Code is cheap; consensus is the new scarce good.
Olares and HAMi: A New Inflection Point for Desktop AI Workstations
HAMi moves from cluster to desktop with Olares.
Every Nation Begins with Textiles
From Anji bamboo weaving to industrialization
Why GPUs Became the Foundation of AI: A GPU Primer for K8s Veterans
A GPU explainer for Kubernetes veterans new to AI. Maps token, model, training, inference, Transformer, Tensor Core, HBM, and KV cache to concepts you already know.
GPU Utilization Is Breaking: AI Infrastructure Needs a New Definition of Efficiency
From GPU utilization to productive GPU-hours.
When an Agent Becomes a Distributed State Machine: Agentic AI Infrastructure Reliability
A practical AI Infra review of Agentic AI reliability, covering a five-dimension framework, fault tolerance, recovery, observability, and hybrid architecture design.
From GPU to Token: The 8-Layer Observability Stack for AI Infrastructure
From GPU hardware, Kubernetes scheduling, inference engines to token cost — understanding the 8-layer observability architecture for modern AI infrastructure.
My Personal AI Stack: Building a Continuously Running Personal AI Infrastructure for ~$100/Month
How I built a personal AI infrastructure using ChatGPT, OpenClaw, Obsidian, GitHub, Lark, GLM-5.1, and a Mac mini M4.
Token Is More Than a Billing Unit, It's Becoming the Resource Unit of the AI Era
The Linux Foundation's Tokenomics Foundation signals a shift: tokens are becoming a core resource in the AI era, much like CPUs in the cloud era.
AI Native Landscape Launches as a Standalone Site
AI Native Landscape has moved to landscape.jimmysong.io with 600+ curated open-source projects, AI skill search support, and a call for community contributions.
Kubernetes as the GPU Control Plane: HAMi v2.9 and Next-Gen AI Infra
Observations on the evolution of AI infrastructure control planes, focusing on HAMi v2.9, GPU scheduling, and Kubernetes resource models.
Kubernetes's Anxiety and Rebirth in the AI Wave
At KubeCon EU 2026, I witnessed Kubernetes' anxiety and transformation in the AI era. This article explores the challenges and future opportunities for Kubernetes in the age of AI.
Day One in Amsterdam: Kubernetes Is Rethinking AI
KubeCon Europe 2026 Day One: How Kubernetes is adapting to the AI infrastructure wave and the evolution of the GPU resource layer.
HAMi Website Refactor: Why HAMi Docs and Website Underwent a Complete Redesign
A systematic upgrade to HAMi’s website and docs, improving community visibility, content structure, search, and usability.
GTC 2026 Eve: AI is Becoming the New Infrastructure
On the eve of GTC 2026, rethinking whether AI is becoming the new infrastructure from NVIDIA's AI Five-Layer Cake, the rise of agent runtime, to AI-native infrastructure.
When GPUs Move Toward Open Scheduling: Structural Shifts in AI Native Infrastructure
A CTO/VP view on open GPU scheduling: CDI, Kubernetes DRA, virtualization data planes, ecosystem governance, and lock-in risk.
AI Learning Resources: 44 Curated Collections from Our Cleanup
A curated collection of AI learning resources we removed from the AI Resources list: awesome lists, courses, tutorials, and cookbooks. These educational materials deserve their own spotlight.
Standing on Giants' Shoulders: The Traditional Infrastructure Powering Modern AI
Before ChatGPT and TensorFlow, there was Hadoop, Kafka, and Kubernetes. This post honors the traditional open source infrastructure that became the foundation of today's AI revolution.
My First Month at Dynamia: Why AI Native Infra Is Worth the Investment
Observations from my first month at Dynamia: From cloud native to AI Native Infra, why this direction is worth investing in, and the key issues and opportunities in compute governance.
The True Inflection Point of ADD: When Spec Becomes the Core Asset of AI-Era Software
Exploring how Spec becomes the governable core asset in Agent-Driven Development (ADD) and the trend toward control-plane engineering systems.
AI Voice Dictation Input Methods Are Becoming the New Shortcut Key for the Programming Era
Comparing Miaoyan, Zhipu, and Shandianshuo voice input methods for developers: speed, stability, command capabilities, and cost models.
From Spatial Data to AI Open Source: Technical Standards, Data Sovereignty, and the Global Divide
How technical standards and data sovereignty shape AI open source paths and infrastructure competition in the global AI era.
Joining Dynamia: Embarking on a New Journey in AI Native Infrastructure
Joining Dynamia as Open Source Ecosystem VP to drive AI-native infrastructure ecosystem development, transforming compute from hardware consumption to core asset.
Running Parallel AI Agents on My Mac: Hands-On with Verdent's Standalone App
A hands-on experience with Verdent's standalone Mac app, exploring how parallel AI agents, isolated workspaces, and task-oriented workflows change real-world development.
2025 Annual Review: The Transformation Journey from Cloud Native to AI Native
A look back at the major changes in 2025: shifting from Cloud Native to AI Native Infrastructure, AI tool ecosystem, and major website improvements.
The Butterfly Effect After Manus Was Acquired by Meta
Manus's acquisition by Meta sparked polarized opinions. This article explores the butterfly effect in AI applications and key lessons for entrepreneurs on growth strategies.
AI Infra Open Source in China: Analysis of Beijing and Shanghai's Plans
Beijing and Shanghai's open source plans reveal opportunities and challenges for China's AI infrastructure, balancing technology and governance.
From 2025 Onwards, Software Engineering Shifts from Code-Centric to Runtime and Cost-Centric
In 2025, software engineering shifts from code-centric to runtime and cost governance. AI and Agents move complexity to runtime, compute, and budget layers, reshaping engineering value.
From Cloud Native to AI Native: Why Kubernetes Is the Foundation for Next-Gen AI Agents
Explores why AI Agents need Kubernetes infrastructure and how Agent orchestration, MCP services, and AI gateways enable production-ready AI architectures.
AI Open Source Landscape: A One-Stop Guide to AI Project Navigation and Scoring System
Comprehensive introduction to the AI Open Source Landscape's positioning, interface, scoring model, and data mechanisms to help developers efficiently discover quality AI projects.
AI 2026: Infrastructure, Agents, and the Next Cloud-Native Shift
2026 AI's turning point: not models, but infrastructure, agentic runtimes, GPU efficiency, and new organizational forms.
What I Saw at COSCon'25: The Real State of Open Source in China
From an engineering and organizer's perspective, real changes at COSCon'25: AI as the default backdrop, discussions returning to engineering issues, and Chinese open source entering a long-term phase.
Decoding Goose: Why It Joined AAIF and What This Means for Agentic Runtime
An analysis of Block's Goose project, why it became one of the first Agentic AI Foundation (AAIF) projects, and what this means for Agentic Runtime and the evolution of AI-Native infrastructure.
ARK: Multi-Agent Systems Are Finally Entering the Engineer's World
How ARK uses cloud-native architecture and declarative runtime to drive engineering adoption of multi-agent systems and shape the Agentic Runtime ecosystem.
Can Open Source Suddenly Disappear? An AI Chat Dev Tool Went 404 Overnight
Lunary, an open-source project in the AI DevTool space, suddenly deleted its GitHub repo, exposing the instability of commercial open source projects.
CNCF in the AI Native Era? The Agentic AI Foundation Is Officially Established
An analysis of the background, strategic urgency, differences and division of labor between Agentic AI Foundation (AAIF) and CNCF/CNAI, and its significance for the AI Native era.
Bun Acquired by Anthropic: A Structural Signal for AI-Native Runtimes
Bun's acquisition by Anthropic marks the first time a general-purpose language runtime is integrated into a large model engineering system, revealing a structural trend for AI-native runtimes.
Agentic Runtime Realism: Insights from McKinsey Ark on 2026 Infrastructure Trends
Analyzing Ark from architecture, semantics, community activity, and engineering paradigms to reveal its impact on 2026 AI Infra trends and the ArkSphere community.
In-Depth Analysis of Ark: Kubernetes for the AI Era or a New Engineering Paradigm Shift?
Analysis of McKinsey's Ark project: architecture, CRDs, control plane, design paradigms, production readiness, and implications for ArkSphere and AI infrastructure.
From Using AI to Relying on AI: Why the Era of AI Engineering Has Yet to Begin
AI's real turning point is moving from using AI tools to building AI systems. Why the era of AI engineering hasn't begun, and the developer opportunity in the next three years.
Antigravity VS Code Setup Guide: Build a Practical AI IDE Workflow
A practical Antigravity setup guide for developers who want a VS Code-style AI IDE, including marketplace switch, AMP and CodeX installation, and workflow tuning.
Cloudflare November 18 Global Outage: The Dangers of Implicit Assumptions in Modern Infrastructure
An analysis of the Cloudflare global outage on November 18, 2025, exploring implicit assumptions, automated configuration pipelines, and systemic risks in modern infrastructure.
The Second Half of Cloud Native: The Era of AI Native Platform Engineering Has Arrived
A decade of cloud native evolution, a look ahead to AI-Native Platform engineering, technical layers, and key changes. KubeCon NA 2025 signals a new era.
NotebookLM: My Most Recommended AI Tool for Learning and Knowledge Organization
Based on months of deep usage, this article analyzes how NotebookLM helps me learn new technologies, read complex documents, generate teaching outlines, and shares future improvement expectations.
Helm v4: Paradigm Convergence and Plugin System Rebuild
An analysis of Helm 4's core changes, including Server-Side Apply, WASM plugin system, kstatus status model, reproducible builds, and content hash caching, with a timeline review of Helm's history.
Kimi K2 Thinking: The True Awakening of China's Thinking Model
Kimi K2 Thinking's open source marks China's entry into thinking models. This article reviews its technical approach and compares it with Claude and Gemini.
Beyond Gateway: Inference Traffic Control Practice with Gateway API Inference Extension
Exploring how Gateway API Inference Extension brings model-aware inference traffic control through InferencePool, InferenceObjective, and metrics-driven routing.
TRAE SOLO vs VS Code: Rethinking Coding Tools from the Perspective of AI Engineering Entities
A comparison of TRAE SOLO and VS Code (Copilot, Agent HQ) via the AI Engineering Entity framework, focusing on automation, collaboration, model transparency, and engineering roles.
Closed-Source Flagships Accelerate, Open-Source Ecosystem Forced to 'Synchronize'
Analysis of closed-source model acceleration and open-source ecosystem response, exploring core engineering contradictions and infrastructure evolution.