Re: Proposal for a Kubernetes AI Conformance Working Group

[Yuan Tang]

Adding WG Serving and WG Batch since some requirements are directly related.

On Monday, July 7, 2025 at 5:39:50 PM UTC-4 Laura Santamaria wrote:

Just catching up on this after the US holidays. +1 to this, and please include me in the discussions.

On Thursday, July 3, 2025 at 8:00:16 AM UTC-5 Camila Macedo wrote:

+1

CAMILA MACEDO

Principal Software Engineer 

RED HAT Operator framework

Red Hat UK

She / Her / Hers

IM: cmacedo

I respect your work-life balance. Therefore, you do not need to answer this email outside of your office hours.

On Wed, Jul 2, 2025 at 11:53 AM ankur parashar pandey <ankurpara...@gmail.com> wrote:

Add me to discussions or slack channel.

On Wed, Jul 2, 2025, 2:42 AM ankur parashar pandey <ankurpara...@gmail.com> wrote:

+1.

On Sat, Jun 28, 2025, 7:23 AM 'Federico Bongiovanni' via dev <d...@kubernetes.io> wrote:

Dear Kubernetes Community,

following up after the SIG Architecture meeting on 6/26/25 [notes] and in accordance with the Kubernetes Working Group Formation guidelines, we are proposing the formation of a new Working Group focused on establishing a Kubernetes AI Conformance certification. The goal of this group is to define a standardized set of capabilities, APIs, and configurations that a Kubernetes cluster must offer to reliably and efficiently run AI/ML workloads. This initiative aims to simplify AI/ML operations on Kubernetes, accelerate adoption, guarantee interoperability and portability for AI workloads, and enable ecosystem growth on an industry-standard foundation.

Below, we address key questions regarding the proposed Working Group:

1. What is the exact problem this group is trying to solve? The core problem is the current lack of a standardized and certified baseline for Kubernetes clusters specifically tailored for AI/ML workloads. While Kubernetes is widely used, reliably and efficiently running AI/ML often requires additional capabilities, specific APIs, and particular configurations beyond standard CNCF Kubernetes Conformance. This leads to fragmentation, interoperability challenges, and increased effort for users to ensure their AI/ML workloads are truly portable across different Kubernetes environments.

2. What is the artifact that this group will deliver, and to whom? The primary artifact will be the Kubernetes AI Conformance specification and a suite of tests to demonstrate conformance. This specification will detail the additional capabilities, APIs, and configurations necessary for AI/ML workloads. It will be delivered to the broader Kubernetes community, including platform providers, distribution maintainers, and end-users who operate or plan to operate AI/ML workloads on Kubernetes. Ultimately, this will enable AI conformance certification.

3. How does the group know when the problem solving process is completed, and it is time for the Working Group to dissolve? The Working Group will consider its primary problem-solving objective complete upon the successful definition and initial adoption of a stable Kubernetes AI Conformance specification. This includes ensuring the conformance significantly simplifies the deployment and management of AI/ML workloads, reduces the need for "DIY" solutions and framework-specific patches, and fosters portability. Once the foundational conformance is established and widely recognized, the ongoing maintenance and evolution of the conformance would be evaluated, and could ideally transition to an existing or newly formed Special Interest Group (SIG) with a long-term charter, at which point the Working Group would dissolve.

4. Who are all of the stakeholder SIGs involved in this problem this group is trying to solve? We propose that this WG starts initially under SIG Architecture. More broadly, this initiative touches upon several core areas of Kubernetes, requiring collaboration with various SIGs like:

• SIG Node: For discovery, allocation, and management of specialized AI hardware (e.g., GPUs, NPUs), and ensuring node health relevant to AI workloads.

• SIG Scheduling: For AI-specific extensions to scheduling logic, gang scheduling, and efficient placement of AI workloads onto nodes with accelerators.

• SIG Storage: For specific storage configurations and capabilities crucial for AI/ML datasets and models.

• SIG Network: For specific networking requirements and high-throughput communication patterns of distributed AI workloads.

• SIG Instrumentation: For defining and ensuring robust observability and telemetry needs of AI/ML workloads.

• SIG Security: For security considerations pertinent to AI/ML environments, including data access, model integrity, and supply chain security.

• SIG Release & SIG Testing: For considerations around the stability and maintenance of a conformant AI Kubernetes environment, and for the conformance testing and release process of the specification.

• SIG Apps: For considerations related to AI application deployment, lifecycle management, and operator patterns.

• SIG API Machinery: For defining and extending Kubernetes APIs to better support AI frameworks and operators.

• SIG Architecture: For ensuring core Kubernetes extensibility points function effectively for common AI operator patterns.

• SIG Auth: For authentication and authorization concerns within AI/ML contexts.

5. What are the meeting mechanics (frequency, duration, roles)? We propose the following initial meeting mechanics:

• Frequency: Weekly or bi-weekly.

• Duration: 60 minutes.

• Roles: The Working Group will have designated Chair(s) responsible for guiding discussions and ensuring progress. A note-taker will be assigned for each meeting, and active participation from all contributors will be encouraged. Agendas and meeting notes will be publicly accessible.

• Communication channels:

  • Slack: a new #wg-ai-conformance will be established for discussions

  • Mailing: a new wg-ai-conformance mailing list will be created and used for official announcements and design discussions. 

6. Does the goal of the Working Group represent the needs of the project as a whole, or is it focused on the interests of a narrow set of contributors or companies? The goal of this Working Group is fundamentally aligned with the broader interests of the Kubernetes project and its community. By simplifying AI/ML on Kubernetes, accelerating adoption, and guaranteeing interoperability and portability, we aim to expand Kubernetes' utility and reach into a critical and growing workload domain. This initiative is designed to benefit all users and contributors by providing a robust, standardized foundation for AI/ML, not just a narrow set of companies or individual contributors.

7. Who will chair the group, and ensure it continues to meet these requirements? The initial organizers of this proposal are prepared to serve as interim chairs to launch and guide the Working Group:

• Mario Fahlandt (https://github.com/mfahlandt)

• Janet Kuo (https://github.com/janetkuo)

We are committed to operating transparently and will adhere to Kubernetes community guidelines to ensure the group continually meets its stated goals and requirements. 

8. Is diversity well-represented in the Working Group? We are committed to fostering a diverse and inclusive environment within this Working Group. We will actively encourage participation from individuals across various companies, geographical locations, technical backgrounds, and experience levels to ensure a wide range of perspectives are represented in the development of the AI Conformance specification.

We believe that a dedicated Working Group is essential to drive the Kubernetes AI Conformance forward effectively and collaboratively. We look forward to discussing this proposal with the community and welcoming contributions.

If something is missing from the people that have been involved in these discussions so far please feel free to add or to ask for clarifications,

Thanks

Federico Bongiovanni (https://github.com/fedebongio) (on behalf of the group in meeting yesterday)

--
You received this message because you are subscribed to the Google Groups "dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dev+uns...@kubernetes.io.
To view this discussion visit https://groups.google.com/a/kubernetes.io/d/msgid/dev/CAPS6%2B43nxFA3oXHg5vWZKJMfCeNYYp0s59kwRwkNC5CMBBrFwg%40mail.gmail.com.

--
You received this message because you are subscribed to the Google Groups "dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dev+uns...@kubernetes.io.

To view this discussion visit https://groups.google.com/a/kubernetes.io/d/msgid/dev/CAHEefoLoHAnd%3DJ1nTR0qikub-9UfR1ReAf_rKomgriLGgqpAeQ%40mail.gmail.com.

[Federico Bongiovanni]

Thank you all for the responses and enthusiasm. I'll be moving to the next step in the next days according to the procedure, formalizing the WG, meetings, communication channels, etc.

On Mon, Jul 7, 2025 at 5:47 PM Elieser Jose Pereira Reyes <elieser....@gmail.com> wrote:

+1 on this. Please include me in any channel or discussion.

To view this discussion visit https://groups.google.com/a/kubernetes.io/d/msgid/dev/163afa17-2ef4-4f95-93d3-da186e2ea129n%40kubernetes.io.

[Bowei Du]

Please include me and liu...@google.com (networking)

Thanks

Bowei

On Wed, Jul 9, 2025 at 11:44 AM 'Federico Bongiovanni' via dev <d...@kubernetes.io> wrote:

Hi all, following the process, submitted this PR to formalize the WG.

Given the traction that the proposal is having, ideally if things move fast we could start meeting next week, that's our goal as of today. Will keep this thread posted.

Kind regards

Fede

On Tue, Jul 8, 2025 at 6:39 AM Francisco Arceo <arceofr...@gmail.com> wrote:

+1 excited to collaborate from the Kubeflow community. :) 

On Tue, Jul 8, 2025 at 9:17 AM Elieser Jose Pereira Reyes <elieser....@gmail.com> wrote:

+1 to this. Happy to collaborate, please include me on any channel or discussion

To unsubscribe from this group and stop receiving emails from it, send an email to wg-batch+u...@kubernetes.io.

--

Francisco Javier Arceo

franciscojavierarceo.github.io 

--
You received this message because you are subscribed to the Google Groups "dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dev+uns...@kubernetes.io.

To view this discussion visit https://groups.google.com/a/kubernetes.io/d/msgid/dev/CAPS6%2B40peqwPiZcTqCKHuMZ3qFHZwec-NAxLwJMi_v2Dg%2B9NpA%40mail.gmail.com.

[Federico Bongiovanni]

Hi all, following the process, submitted this PR to formalize the WG.

Given the traction that the proposal is having, ideally if things move fast we could start meeting next week, that's our goal as of today. Will keep this thread posted.

Kind regards

Fede

On Tue, Jul 8, 2025 at 6:39 AM Francisco Arceo <arceofr...@gmail.com> wrote:

+1 excited to collaborate from the Kubeflow community. :) 

On Tue, Jul 8, 2025 at 9:17 AM Elieser Jose Pereira Reyes <elieser....@gmail.com> wrote:

+1 to this. Happy to collaborate, please include me on any channel or discussion

El lunes, 7 de julio de 2025 a las 21:37:42 UTC-3, Yuan Tang escribió:

To unsubscribe from this group and stop receiving emails from it, send an email to wg-batch+u...@kubernetes.io.

[Keith Mattix]

I'm also looking forward to joining from the networking and Gateway API perspective! Excited to work with everyone

[ANDRE ALMAR]

+1 

--

Andre Almar

Lead Infrastructure Engineer - Thoughtworks AI Labs

Email

andre...@thoughtworks.com

[Dawn Chen]

Please include +Sergey Kanzhelev from SIG Node. There might be more work from the node perspective, and we will identify more folks once the work is scoped out. Thanks!

You received this message because you are subscribed to the Google Groups "wg-serving" group.
To unsubscribe from this group and stop receiving emails from it, send an email to wg-serving+...@kubernetes.io.
To view this discussion visit https://groups.google.com/a/kubernetes.io/d/msgid/wg-serving/25145e65-ad73-4fe6-9813-d9bc844e7b1bn%40kubernetes.io.

[Sudhanshu Prajapati]

+1 to this, please include me in the discussions.

-- 

Sudhanshu Prajapati | Senior Developer Advocate

Improving – It’s what we do.™

improving.com
Software Development | Consulting Services | Training & Coaching | Outsourcing | Community

To view this discussion visit https://groups.google.com/a/kubernetes.io/d/msgid/wg-serving/CAJo%3DGk6C36Cvzw7ueqeiWvzVEkPVNmTxv-oVT%2B3tM3gx0axVUA%40mail.gmail.com.

[Rithwik Krishna]

+1 , would love to join.

[Jonathan Innis]

+1, would love to join from the SIG autoscaling side.

Jonathan

[Vishwa Gandhi]

+1

I would like to join as well.

[rigin rajan]

+ 1

I would like to join

Regards

Rigin Rajan

[debabrata sarkar]

+1

I would like to join as well.

Regards,

Deb

On 25 Jul 2025, at 23:22, rigin rajan <rigi...@gmail.com> wrote:



To view this discussion visit https://groups.google.com/a/kubernetes.io/d/msgid/wg-serving/CAPFXqLobJdjeKNzpvS5voVdoiM83PUw_iJSbRD1Xr%2B0CYACkTQ%40mail.gmail.com.

[Laura Santamaria]

The working group has been formed [1], and meetings are set [2] for Thursdays at 10a Pacific. Join the channel in Kubernetes Slack [3] and the mailing list [4] for more collaboration :)

Cheers,

Laura

[1]: https://github.com/kubernetes/community/pull/8515

[2]: https://github.com/kubernetes/community/blob/master/wg-ai-conformance/README.md#meetings

[3]: https://kubernetes.slack.com/messages/wg-ai-conformance

[4]: https://groups.google.com/a/kubernetes.io/g/wg-ai-conformance

--

Laura Santamaria (she/her)

https://linktr.ee/nimbinatus

On Thu, Jul 31, 2025 at 8:45 AM Ayman <mohammeda...@gmail.com> wrote:

+1 to this. Would love to join

To view this discussion visit https://groups.google.com/a/kubernetes.io/d/msgid/dev/C428E0CB-6BD3-4D34-86D1-5890E4F7D90A%40gmail.com.

[Ayman]

+1 to this. Would love to join

To view this discussion visit https://groups.google.com/a/kubernetes.io/d/msgid/dev/C428E0CB-6BD3-4D34-86D1-5890E4F7D90A%40gmail.com.

[caio cesar]

Please also add me.

Caio Davi

[Davanum Srinivas]

Folks,

PLEASE stop posting "me too" now that
- there's a slack

- there's a meeting setup

- there's a mailing list

https://github.com/kubernetes/community/blob/master/wg-ai-conformance/README.md#contact

Thanks,

Dims

PS: will reject any more posts that will end up in dev@ mailing list.

On Fri, Aug 1, 2025 at 2:14 PM Abhinandh B G <abhin...@gmail.com> wrote:

pls add me as well

Thanks
Abhinandh (Abhi)

--

Davanum Srinivas :: https://twitter.com/dims

[Abhinandh B G]

pls add me as well

Thanks
Abhinandh (Abhi)

On Thursday, July 31, 2025 at 8:56:15 PM UTC+5:30 caio cesar wrote:

[Kunle Babajide]

Please add me on as well.

[Tu Xudong]

Hi, everyone 

First, i would like to join.

Second, we're working on

1)  broadcast this import trend and problem to relative people, see https://github.com/turtacn/PersonalAI/blob/master/community/CNCF/Kubernetes/01_CNCF-Kubernetes-AI-Conformance.md

2) technical insights on reference architecture on  agentic ai infra. & platform , see https://github.com/turtacn/PersonalAI/blob/master/examples/trends/04_agentic-ai-reference-architecture%26platform.md

3) proposal on this topic, see https://github.com/turtacn/PersonalAI/blob/master/community/CNCF/Kubernetes/02_Proposal-of-CNCF-SIG-on-Kubernetes-AI-Conformance-AAI-perspecitve.md  && https://github.com/turtacn/PersonalAI/blob/master/community/CNCF/Kubernetes/02_Proposal-of-CNCF-SIG-on-Kubernetes-AI-Conformance-AAI-perspecitve-CN.md

Best, 

xudong