DevSecOps firm Jozu is leading two CNCF-backed open source projects, KitOps and ModelPack, to make AI/ML model delivery reproducible, verifiable, and as portable as containers.
DevSecOps company Jozu is driving a new era of open source standardisation in AI and ML through its leadership of two CNCF-backed projects, KitOps and ModelPack. Both initiatives aim to simplify and standardise how AI/ML models are packaged, versioned, and deployed across environments.
KitOps is an open source DevOps tool that packages an AI or ML model, dataset, codebase, and configuration into a reproducible artefact known as a ModelKit. It enables cryptographically verifiable, compliant, and reproducible deployments while integrating seamlessly with existing container registries and Kubernetes infrastructure, requiring no additional tooling.
ModelPack, on the other hand, defines open standards for packaging, distributing, and running AI artefacts in cloud-native environments. By extending the OCI (1.1) standard, it supports the large artefacts common in AI—ranging from models and datasets to complex applications—bringing container-like portability to AI workflows.
Both projects, supported by the Cloud Native Computing Foundation (CNCF), are recognised as the first to deliver such portability, ensuring reproducibility and security for AI/ML workflows.
“The AI/ML ecosystem has lacked open standards for packaging, versioning and sharing projects,” said Mohamed Nanabhay, Managing Partner at Mozilla Ventures. Görkem Ercan, CTO and Co-founder of Jozu, added, “ModelKits make AI deployments cryptographically verifiable, compliant and reproducible across any Kubernetes environment.” Andrew Block, ModelPack Maintainer at Red Hat, highlighted the community’s collaborative potential to drive innovation.
Jozu now offers enterprise support for KitOps and ModelPack through its Jozu Hub platform, enabling full DevSecOps management of ModelKit projects across connected, on-premises, and air-gapped environments.
