Rise Edge: Edge AI Compute Platform
Edge Cloud-Native · Central Train, Edge Infer · 6 Unifications
Platform Overview
Resource/App/Scheduling/Auth/Monitor/Ecosystem
Node onboarding & model updates
Managed & dedicated clusters
Edge chip vendors supported
Unified Resource Management
Multi-Runtime Support
Compatible with both KubeEdge and OpenYurt edge runtimes. Manage clusters with different runtimes on a single platform, no vendor lock-in.
Managed & Dedicated Modes
Managed mode (Virtual Edge Cluster) for lightweight onboarding; Dedicated mode for independent operations. Unified gateway manages both.
Edge vGPU Slicing
Powered by Rise VAST vGPU engine, fine-grained GPU/NPU partitioning on edge nodes. Multiple inference services share one edge card for dramatically improved utilization.
Heterogeneous Chip Support
Supports NVIDIA Jetson/T4, Ascend 310P/910B, KunlunXin, Hygon and more edge AI accelerators, unified across x86 and ARM architectures.
OTA Node Onboarding
Batch node registration, automatic dependency installation, auto-upgrade, and backup/restore. Select container engine and edge runtime, one-click onboarding with zero config errors.
Unified Edge Gateway
edge-gateway unifies cloud-edge API access, communication, and OTA control channels, abstracting away underlying network complexity.
Application & Model Distribution
Edge App Store
Unified app/model repository supporting containers, models, agents, and workflows with full lifecycle management. Visual deployment, one-click push, canary upgrades.
OTA Model Delivery
One-click push trained models to designated edge node groups. Canary releases, batch rollouts, incremental delta updates, zero-downtime hot updates.
Version Management & Rollback
Complete edge model and application version tracking with one-click rollback to any stable version, ensuring edge business continuity.
Offline Autonomy
KubeEdge / OpenYurt edge autonomy ensures inference services continue during network disconnection, with automatic state sync on reconnection.
Cloud Upstream — optional
Distributed Edge Sites
Visual defect detection
Traffic analysis
Customer flow·offline
Behavior analysis
Unified Operations & Access Control
NodeGroup Resource Management
Manage physical resources by node groups with multi-tier structure, resource authorization, and least-privilege node access. Dual mapping between resource tenants and application tenants.
Multi-cluster Management
Cross-region multi-cluster onboarding compatible with different K8s distributions. Single console for all edge site cluster status, node resources, and container workloads.
Edge Monitoring & Alerting
Real-time collection of edge GPU/NPU utilization, temperature, power, and inference latency metrics. Tiered alerting (email/SMS/DingTalk/WeCom) with on-call rotation.
Remote Diagnostics
View edge node logs and run diagnostic commands from the cloud, with remote terminal access. No on-site maintenance needed.
Auto-healing
Automatic edge node fault detection, inference service migration to healthy nodes in the same group, with faulty node isolation and reporting.
Security & Audit
End-to-end encrypted cloud-edge communication, strict inter-container resource isolation, full traceability of operations logs and resource changes for industrial compliance.
Use Cases
Smart Manufacturing QA
Deploy visual inspection models to production line edge nodes for millisecond-level defect detection. Model iterations pushed via OTA with zero downtime.
Intelligent Transportation
Roadside edge nodes run vehicle recognition and traffic flow analysis locally, avoiding backhaul latency. Supports Ascend 310P edge cards.
Smart Retail
In-store edge devices run customer flow analysis and product recognition. Offline-capable with cloud-managed model and app versions across hundreds of stores.
Smart Campus & Security
Campus cameras connected to edge AI nodes for local behavior analysis and alerting, reducing bandwidth costs and keeping data on-premises.