CIOPages
DirectoryCloud Infrastructurehami

hami

Open SourceFunded

Middleware for managing heterogeneous AI devices in Kubernetes environments

Visit Website

About hami

HAMi is an open-source middleware designed to manage heterogeneous AI computing devices such as GPUs, NPUs, and other accelerators within Kubernetes clusters. It enables efficient device virtualization, sharing, and resource isolation, allowing multiple pods to utilize AI hardware resources dynamically without requiring changes to existing applications. HAMi supports a wide range of devices including NVIDIA GPUs, Cambricon MLU, HYGON DCU, and Huawei Ascend NPUs, providing a unified interface for heterogeneous device management.

Targeted at enterprises operating in cloud infrastructure environments, HAMi is widely adopted across industries like finance, telecommunications, manufacturing, and education. It enhances scheduling decisions by considering device topology and policies, optimizing resource utilization in both public and private cloud deployments. As a CNCF sandbox project, HAMi benefits from an active community of contributors and users, ensuring continuous development and integration with Kubernetes orchestration. Its monitoring capabilities and web UI support operational visibility and management at scale.

Key Capabilities

  • Heterogeneous device virtualization and management
  • Dynamic device sharing and resource isolation
  • Unified interface for multiple AI hardware types
  • Topology-aware scheduling for Kubernetes pods
  • Integrated monitoring and web-based UI

Integrations

KubernetesNVIDIA device pluginHelm package manager

This profile was compiled by CIOPages from public sources with AI assistance, and may be incomplete or out of date. It is informational only and not an endorsement. Represent this vendor? or .

Quick Facts

project-hami.github.io/HAMi
CategoryCloud Infrastructure
PricingOpen Source
DeploymentOpen Source, On-Premises, Cloud
Target SizeEnterprise