Skip to content

[Sandbox] AI Runway #488

@squillace

Description

@squillace

Project summary

AI Runway is an abstract set of operators that enable easier deployments of a wide array of AI inferencing engines and models in Kubernetes.

Project description

AI Runway gives you a web UI and a unified Kubernetes CRD (ModelDeployment) to deploy models across multiple inference providers. Browse HuggingFace, pick a model, click deploy. The intention is to make it easier for the Kubernetes and cloud native ecosystem to deploy and learn about deploying various combinations of open source inference engines and models in Kubernetes.

At the moment, there are a wide array of inference engines and servers to host an inference endpoint, but each one requires specific installation process and mastering the very different details of each. This choice is a beautiful aspect of the ecosystem, but it also makes it harder to start building agentic and other applications in Kubernetes that use inference endpoints of various kinds until you've mastered the stack choice and the engine you end up choosing.

AI Runway helps break that logjam, getting deployments started easily with the widest possible array of choices from the ecosystem. The provider extension model means that AI Runway can add any current or future inference engine without modifying the deployment abstraction, and can be used locally or in any cloud native Kubernetes distribution. If successful, the project might serve as a useful exercise implementation for a future Kubernetes AI Inference API proposal.

Org repo URL (provide if all repos under the org are in scope of the application)

N/A

Project repo URL in scope of application

https://github.com/kaito-project/airunway

Additional repos in scope of the application

No response

Website URL

https://github.com/kaito-project/airunway

Roadmap

https://github.com/kaito-project/airunway/issues

Roadmap context

We are building a project page and will update the application as soon as it is completed.

Contributing guide

https://github.com/kaito-project/airunway/blob/main/CONTRIBUTING.md

Code of Conduct (CoC)

https://github.com/kaito-project/airunway/blob/main/CODE_OF_CONDUCT.md

Adopters

No response

Maintainers file

https://github.com/kaito-project/airunway/graphs/contributors?from=2%2F14%2F2026

Security policy file

kaito-project/airunway#279

Standard or specification?

N/A, though this is likely to change.

Business product or service to project separation

This is not part of any business product or service.

Why CNCF?

The CNCF is the open source foundation that houses the finest in cloud native projects the world can use forever. This project makes it possible to use with some ease the widest array of models with a wide array of stacks in a fully vendor-neutral, Kubernetes-native experience -- something that belongs in the CNCF.

Benefit to the landscape

There are no other similar projects in the CNCF at the moment, though we feel sure they will appear. The LF has a large array of projects in its AI organizations, but none of them are really aimed at the same market.

Cloud native 'fit'

it logically fits in https://landscape.cncf.io/?group=ai-native&view-mode=grid here, possibly in "AI Native Infra" or in Application Definition in the core map. As it "assembles" other projects into a deployment, possible "framework" might make sense......

Cloud native 'integration'

It currently can make use of Kaito, llm-d, vllm, ray, slurm, etc..... any part of the system can be added to the tool as a provider independently of any other. it is a true "integration" piece for CNCF AI tooling.

Cloud native overlap

it really doesn't, in that it's an integration and usability tool for other projects.

Similar projects

N/A

Landscape

Not yet.

Insights

Not yet.

Trademark and accounts

  • If the project is accepted, I agree to donate all project trademarks and accounts to the CNCF

IP policy

  • If the project is accepted, I agree the project will follow the CNCF IP Policy

Will the project require a license exception?

No.

Project "Domain Technical Review"

Not yet.

Application contact email(s)

ralph.squillace@microsoft.com,seozerca@microsoft.com,david.justice@microsoft.com

Contributing or sponsoring entity signatory information

If an organization:

Name Address Type (e.g., Delaware corporation) Signatory name and title Email address
Microsoft Corporation One Redmond Way, Redmond WA 98052 US corporation Brendan Burns -- Corporate Vice President and Technical Fellow bburns@microsoft.com

Or, if an individual or individual(s):

Name Country Email address

CNCF contacts

No response

Additional information

No response

Metadata

Metadata

Type

No type
No fields configured for issues without a type.

Projects

Status

🏗 Upcoming

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions