Project HAMi

HAMi is a @CloudNativeFdn open-source GPU virtualization middleware for sharing, isolating, and scheduling heterogeneous accelerators on Kubernetes.

[Project HAMi @HAMiProject]

HAMi is coming to KubeCon China 2026!

Dynamic MIG, GPU virtualization at scale, and heterogeneous AI infrastructure, see what the HAMi community will be sharing in Shanghai.

https://t.co/TsCYWVd1nr

#HAMi #KubeCon #AIInfra

[Project HAMi @HAMiProject]

HAMi v2.10.0 is released.
New features include Flexible MIG, composable scheduling strategies, broader accelerator support, and deeper Kubernetes ecosystem integrations.
Learn more:
https://t.co/Uru8LqQ0CV
#HAMi #Kubernetes #CloudNative #AIInfrastructure

[Project HAMi @HAMiProject]

GPU sharing is easy to schedule. Enforcing limits is harder.

We tested KAI Scheduler + HAMi-core:
✅ 2 Pods, 1 GPU
✅ Per-Pod quotas
✅ Over-quota allocations fail
KAI schedules. HAMi-core enforces.
https://t.co/3FFY09aKMX

[Project HAMi @HAMiProject]

Reminder: Today is the deadline for project proposals. Remember to submit your Linux Foundation Mentorship mentee application
https://t.co/SvXcg6Sbtg

[Project HAMi @HAMiProject]

🎂 HAMi turns 5!

Five years ago today, HAMi had its first commit and began an open source journey toward making GPU sharing and heterogeneous computing more efficient.

A huge thank you to all contributors, users, and ecosystem partners who have shaped HAMi along the way.

[Project HAMi @HAMiProject]

New HAMi lab: run two KServe Predictor replicas on one NVIDIA GPU with HAMi DRA GPU sharing.
Follow the step-by-step guide:
https://t.co/ZbpFnRo1Zb
#HAMi #KServe #Kubernetes

[Project HAMi @HAMiProject]

HAMi Community Meeting｜August 5, 2026 https://t.co/Pw4RVs1afx via @YouTube

[Project HAMi @HAMiProject]

Want to contribute to open-source GPU infrastructure? HAMi has four projects open for LFX Mentorship 2026 Term 3.

Apply to become a mentee by August 18:
https://t.co/SvXcg6Sbtg

#LFXMentorship #OpenSource

[Project HAMi @HAMiProject]

Meet HAMi at KubeCon + CloudNativeCon Japan 2026 in Yokohama, Jul 28–30. Find us at Booth T-6, Project Pavilion (Pacifico Yokohama 3F).
#KubeConJapan #KubeCon #HAMi

[Project HAMi @HAMiProject]

📷 HAMi Community Meeting | July 22, 2026 Record https://t.co/b09qhNyL5R

[Shivam Kumar @maishivamhoo]

Excellent blog on GPU sharing in Kubernetes with @HAMiProject. It clearly explains Time Slicing, MPS, MIG, and how HAMi enables fine-grained GPU sharing for better utilization. Thanks @SaiyamPathak, Shubham Katara, and @kubesimplify.

https://t.co/5dBeqPVbAm

[Project HAMi @HAMiProject]

We now have an Events page, so go check it out! We'll be maintaining all community events on Google Calendar. https://t.co/hRHfK7B5tg

[Project HAMi @HAMiProject]

Learn how HAMi GPU slicing works with Kueue quota admission, allowing two vGPU jobs to run while keeping excess workload suspended before scheduler-level Pending.
https://t.co/MZgagnj8eA
#Kubernetes #GPU #Kueue #HAMi

[Project HAMi @HAMiProject]

GPUs are expensive. Idle compute is, too.
At vLLM Meetup Shanghai, HAMi author Li Mengxuan shared a practical path from single-node vLLM to PD disaggregation + GPU virtualization.
Recap: https://t.co/fSmLsjLV5x
#vLLM #HAMi

[Project HAMi @HAMiProject]

🚀 Are you really using your compute efficiently?

At the latest vLLM Meetup, HAMi co-creator Mengxuan Li shared the 3 stages of optimizing LLM inference clusters and why GPU utilization is only the beginning.

#HAMi #vLLM #LLM #GPU #AIInfrastructure #Kubernetes

[CNCF @CloudNativeFdn]

HAMi is now a CNCF Incubating Project!

HAMi helps platform teams efficiently share and schedule GPUs and other accelerators on Kubernetes through vendor-neutral GPU virtualization, making AI infrastructure more efficient as workloads scale.

Congratulations to the HAMi team on this milestone! 🚀

Learn more: https://t.co/CC8JvoCC2E

[Project HAMi @HAMiProject]

@CloudNativeFdn Thank you, CNCF! We’re proud of this milestone. If anyone want to explore or contribute, feel free to join our Discord: https://t.co/HHnNQ0wKCz

[Project HAMi @HAMiProject]

🚀 HAMi is heading to WAIC 2026!

Visit us at Booth H2-B401 to explore how open source is making heterogeneous GPU resource management more efficient for AI infrastructure.

Looking forward to meeting developers, users, and ecosystem partners in Shanghai. #WAIC2026

[Project HAMi @HAMiProject]

🎉 The recording of the July 8 #HAMi Community Meeting is now live!

It's also the first video on the new HAMi Community YouTube channel 🚀

📺 https://t.co/TmMtqkv2Gr

Subscribe for future community meetings, demos, and technical updates.

#CNCF #Kubernetes

[Project HAMi @HAMiProject]

New lab from the HAMi community is live! 🎉
Explore how @volcano_sh vGPU works with Gang Scheduling and Queues in this hands-on lab.
https://t.co/4TfOo1RPYZ
#HAMi #Kubernetes #GPU