HAMi at KubeCon + CloudNativeCon Japan 2026
KubeCon + CloudNativeCon Japan 2026 was held July 28-30 at the Pacifico Yokohama. HAMi took part with a main-stage session and a booth in the Project Pavilion. Reza Jelveh presented the session and staffed the booth.
For the full overview, talk slides, and the SNOW case study, see the KubeCon Japan 2026 page.
HAMi in the Keynote
HAMi was featured in the opening keynote, appearing in the slides of CNCF CTO Chris Aniszczyk and Jonathan Bryce, Executive Director, Cloud and Infrastructure, The Linux Foundation.
On Stage: Shared GPU Scheduling + Proactive Autoscaling
The session Shared GPU Scheduling & Proactive Autoscaling: A Production Blueprint for 1000+ GPUs was presented by Jeonghyun Kim (AI Engineer at SNOW) and Reza Jelveh.
The session described how SNOW serves Snow, Epik, and B612, three widely used AI applications reaching more than 200 million users, on over 1000 NVIDIA A100 GPUs. The architecture combines HAMi for GPU sharing with KEDA for proactive autoscaling, halving the GPUs required for the same traffic and improving recovery under load.
The complete metrics, the SNOW case study, and the talk slides are available on the KubeCon Japan page.
At the Booth
At the HAMi booth in the Project Pavilion, engineers stopped by with questions including:
- How to introduce HAMi into an existing cluster
- How HAMi interoperates with Volcano, Kueue, and the NVIDIA KAI Scheduler
- Practical GPU scheduling problems: memory fragmentation, idle GPUs behind batch jobs, and inference workloads unable to obtain a predictable GPU slice
Learn More
The talk, case study, slides, and ways to get involved are collected on the KubeCon Japan 2026 page. To join the discussion, visit the HAMi community.
