Kubeflow
A managed, Kubernetes-based platform on KakaoCloud for building, training, and deploying machine learning workflows, with GPU MIG partitioning and per-namespace access control.
What it does
- Model training
- Model inference
- Model hosting
Features 4
- User permission management (RBAC)RBAC gives each user and group an independent namespace with fine-grained permissions and resource access control.
- GPU MIG instancesSplits one GPU into up to 7 independent instances so multiple jobs can run at the same time.
- Real-time monitoringNamespace resource usage can be checked in real time in the Kakao Cloud console.
- Action logPer-namespace event lookup to understand job status for analysis and monitoring.
Sources
- Pricing
-
Fetched 2026-09-05, HTTP 200. 상품 이용 요금 table: 'Kubeflow 한시적 무료 *VAT 별도'; 기본 과금 항목 lists Kubernetes Engine 클러스터 이용 요금, File Storage 서비스 이용 요금 and 퍼블릭 IP 이용 요금 as what a Kubeflow deployment is actually charged for. The batch had cited the general calculator, which carries no Kubeflow line — consistent with this page, but this page is the source.
- Description
-
Official documentation · 5 Sep 2026
Fetched 2026-09-05, HTTP 200 (kakaocloud.com/services/kubeflow 301s here). Title 'Kubeflow | 카카오클라우드'; the page introduces Kubeflow as '쿠버네티스 환경에서 머신러닝 워크플로우를 쉽고 효율적으로 실행하는 오픈소스 플랫폼 서비스' with a feature list and customer cases.
- Sold within
-
Official documentation · 5 Sep 2026
The product is sold and documented as a KakaoCloud service; the page sits on kakaocloud.com under KakaoCloud's own service catalogue.
- Interfaces
-
Official documentation · 12 Oct 2026
Page: '카카오클라우드 콘솔에서 네임스페이스의 자원 사용량을 실시간으로 확인'.
- Intended for
-
Case studies name enterprise customers Kakao Games, Daiso and ScatterLab; targets ML practitioners running the full model development/training/deployment workflow
Also from Kakao Enterprise
-
AI Insight Developer tool
A GPU monitoring dashboard on KakaoCloud that tracks utilization, memory, temperature, idle ratio, and error metrics for GPU resources across Kubernetes Engine and Virtual Machine environments.
Open Kakao Enterprise in the directory
Something wrong here? Send a correction — quote this product id: kakao-enterprise-kubeflow.