Topic Collection for KubeCon + CloudNativeCon NA 2026

[gaius-qi]

We are preparing for the Dragonfly Maintainer Track at KubeCon + CloudNativeCon North America 2026! We are looking for maintainers or contributors who are interested in presenting on behalf of the Dragonfly community.

How to Apply:
If you are interested, please comment on this issue with the following details:

Draft Title:
Speaker Name(s):
Brief Abstract:

[pmady]

Hi @gaius-qi,

Thanks for the heads-up! I'm very interested in submitting a talk for the KubeCon NA 2026 Maintainer Track. I'll keep an eye on this issue for updates on when submissions open.

Looking forward to representing Dragonfly at KubeCon!

[gaius-qi]

Hi @gaius-qi,

Thanks for the heads-up! I'm very interested in submitting a talk for the KubeCon NA 2026 Maintainer Track. I'll keep an eye on this issue for updates on when submissions open.

Looking forward to representing Dragonfly at KubeCon!

Hi @pmady,

That sounds great! Please provide the draft template about talking.

Draft Title:
Speaker Name(s):
Brief Abstract:

[pmady]

Hi @gaius-qi / @mingcheng ,

I am incredibly excited to represent the community on the Maintainer Track at KubeCon NA!

Before I officially submit a topic, I want to ensure my session aligns perfectly with the organization's goals for this year. I have drafted an abstract focusing on "Accelerating HPC and AI Workloads with P2P Distribution" (which I am happy to run with), but my primary goal is to deliver what the project needs most right now.

Are there specific roadmap items, major upcoming releases, or core adoption use-cases the maintainer team wants to prioritize for the NA audience?

Let me know what narrative serves the Dragonfly ecosystem best right now, and I will tailor the session architecture around that. If you prefer the HPC/AI angle, I will drop the formal abstract below!

Draft Title: Dragonfly in Hyperscale: Accelerating High-Performance Computing (HPC) and AI Workloads with P2P Distribution
Speaker Name(s): Pavan Madduri
Brief Abstract: As High-Performance Computing (HPC) and Agentic AI workloads increasingly dominate Kubernetes ecosystems, platform teams face a critical new bottleneck: distributing massive container images, training datasets, and model weights across thousands of nodes without saturating network bandwidth. In this Maintainer Track session, we will explore how Dragonfly solves this data-gravity challenge at scale. We will dive deep into the architectural principles of Dragonfly's P2P distribution network, showcasing implementations that optimize container registry pressure and drastically reduce pod startup times for data-heavy workloads. Attendees will gain actionable insights into Dragonfly's internal mechanics, recent community updates, and architectural best practices for integrating P2P distribution into modern, high-demand cloud-native platforms.

[pmady]

hey @gaius-qi, i'd like to propose a talk around P2P-based AI model distribution with dragonfly — covering the huggingface/modelscope backends and the injector webhook work. something like:

Draft Title: P2P-Powered AI Model Distribution with Dragonfly
Speaker Name(s): Pavan Mady (@pmady)
Brief Abstract: How Dragonfly's P2P architecture extends beyond container images to distribute large AI/ML models from HuggingFace and ModelScope across Kubernetes clusters, plus zero-config injection via the dragonfly-injector webhook.

happy to adjust if there's a different angle the team would prefer, or if there's a specific topic you'd want covered from the dragonfly side. also open to co-presenting if anyone's interested. lmk!