yue li's Speaker Profile @ Sessionize

Move the lease from etcd to memory

In Kubernetes, node heartbeats are implemented through Lease objects in the coordination.k8s.io API group. By default, each node updates its Lease every 10 seconds to signal liveness.

For example, in a 5,000-node cluster, this results in:
- 500 write operations per second (5,000 nodes ÷ 10 seconds)
- 500 read operations per second (as controllers check these leases)

This sustained load on etcd creates significant:
- Write amplification from frequent small updates
- Storage overhead from maintaining revision history
- Network traffic between etcd members

These factors make etcd a scaling bottleneck for large clusters, particularly for this high-frequency, low-value coordination data.

From Automation to Certification: Our Journey to Kubernetes AI Conformance

In this session, we will share our journey of achieving Kubernetes AI Conformance certification for our AI platform. We will provide a detailed overview of the certification process, including the steps we took to ensure our platform met the rigorous standards defined by WG AI Conformance. In addition, according to WG AI Conformance timelines, automated tests are unavailable for the certification process before Kubernetes v1.37. To address this challenge, we developed an automation tool to streamline the certification process.

We will demonstrate how our automation tool works, highlighting its key features and benefits. Attendees will learn how to leverage this tool to simplify their own certification journeys, saving time and resources. By the end of the session, participants will have a clear understanding of the Kubernetes AI Conformance certification process and how to utilize our automation tool to achieve certification for their AI platforms.

Beyond the Badge: Building an AI-Conformant Kubernetes Platform

As AI workloads move into production, Kubernetes platform vendors face a growing interoperability challenge. Accelerator management, distributed scheduling, inference networking, observability, security, and AI operators often behave differently across platforms, creating friction for both vendors and end users.

In this session, we will share a vendor’s journey toward Kubernetes AI Conformance, from assessing platform readiness and mapping product capabilities to conformance requirements, to identifying gaps and producing verifiable evidence.

We will also discuss what the conformance program validates, what remains outside its scope, and where vendors may encounter ambiguity as the requirements continue to evolve. Attendees will leave with a practical framework for evaluating an existing Kubernetes platform, planning conformance work, and translating certification requirements into measurable portability, reliability, and customer value.

AI-native Govenance and Compliance Operator oriented for Kubernetes

it is an e2e governance and compliance scenarios cross AI BOM, Agent Skills, Image Vulnerabilities, and k8s Cluster Security.

it plan to extend the K8s compliance operator model to LLM workloads. This will leverage existing open-source tools such as AIBOM, skill-scanner, Copacetic, and Kubescape into one practical governance workflow.

It would keep track of AI components, review agent skills, patch vulnerable images, check Kubernetes security policy, and generate the evidence needed for compliance reviews.

AI Agents as Pods: Running Multi-Agent Orchestration Natively on Kubernetes

Most AI agent frameworks run on developer laptops. Gas City, an open-source orchestration SDK, treats Kubernetes as a first-class runtime — running each agent session as a Pod with native client-go integration, configurable resource limits, service accounts, and in-cluster service discovery.

This session presents Gas City's Kubernetes runtime provider. A declarative city.toml defines the agent fleet — providers (Claude Code, Codex, Gemini), resource profiles, and container images — while the controller reconciles desired state against running Pods. Each agent session runs in its own Pod with projected workspace volumes and injected Dolt (Beads database) service endpoints. When an agent's context window fills up, the Pod is recycled and a new one resumes the workflow molecule where the previous session left off.

Building Ultra-Large-Scale Cloud-Native Edge Systems Using Chaos Engineering

Fast growing technologies, such as 5G networks, industrial Internet, and AI, are giving edge computing an important role in driving digital transformation. As each new technology brings benefits, it brings challenges. First, there are massive heterogeneous edge devices and it encompass a broad range of device types. Second, Edge devices are often located in unstable and complex physical and network environments, such as limited bandwidth, high latency, etc. How to overcome these challenges and build a stable, large-scale edge computing platform needs to be resolved.
KubeEdge is an open source edge computing framework that extends the power of kubernetes from central cloud to edge. Now, Kubernetes clusters powered by KubeEdge, can stably support 100,000 edge nodes and manage more than one million pods.
In this session, we will share the Key challenges of manage massive heterogeneous edge nodes and tell how using ChaosMesh to makes KubeEdge more Reliable in large-scale edge nodes.

Performance Testing Benchmark Framework with Locust: Enterprise Best Practices and CI/CD Integration

we will explore how to use the open-source tool Locust to build an efficient product performance testing benchmark framework and integrate it into the enterprise's Continuous Integration/Continuous Deployment (CI/CD) processes. By sharing best practices from enterprise implementations, this topic will guide attendees on how to effectively conduct performance testing in Kubernetes environments to ensure products meet predetermined performance standards before release.

Chaos Engineering practice under ultra-large-scale cloud-native edge computing

Fast growing technologies, such as 5G networks, industrial Internet, and AI, are giving edge computing an important role in driving digital transformation. As each new technology brings benefits, it brings challenges. First, there are massive heterogeneous edge devices and it encompass a broad range of device types. Second, Edge devices are often located in unstable and complex physical and network environments, such as limited bandwidth, high latency, etc. How to overcome these challenges and build a stable, large-scale edge computing platform needs to be resolved.
KubeEdge is an open source edge computing framework that extends the power of kubernetes from central cloud to edge. Now, Kubernetes clusters powered by KubeEdge, can stably support 100,000 edge nodes and manage more than one million pods.
In this session, we will share the Key challenges of manage massive heterogeneous edge nodes and tell how using ChaosMesh to makes KubeEdge more Reliable in large-scale edge nodes.

Enhancing Product Release Quality with Chaos Engineering: Best Practices with Chaos Mesh

Enhancing Product Release Quality with Chaos Engineering: Best Practices with Chaos Mesh in Enterprise CI/CD Pipelines.
This talk will explore how enterprises can leverage Chaos Mesh, an open-source chaos engineering platform, to simulate real-world scenarios and discover hidden issues in development stages. By integrating chaos experiments into CI/CD pipelines, organizations can proactively improve product release quality, ensuring high availability and resilience.
We will dive into practical strategies, real-world examples, and best practices from leading enterprises that have successfully adopted this approach.

Speaker

yue li

Actions

Links

Area of Expertise

Sessions