# RiseUnion (睿思智联) — Complete Reference > Beijing RiseUnion Technology Co., Ltd. specializes in AI heterogeneous computing virtualization and management. The company is the core maintainer of HAMi, a CNCF sandbox project for GPU sharing, scheduling, and resource management in Kubernetes. Website: https://www.theriseunion.com --- ## Products ### Rise VAST — AI Compute Management Platform (HAMi Enterprise Edition) Enterprise-grade software-defined heterogeneous compute pooling platform, co-developed by RiseUnion and 4Paradigm. **Key Metrics:** 10x GPU utilization | 70% cost reduction | 5-minute deployment | 99% hardware compatibility **Features:** - **Compute Pooling:** Software-defined heterogeneous compute resource pooling across GPU types - **vGPU Virtualization:** Fractional GPU sharing with memory and compute isolation - **Intelligent Scheduling:** Priority-based scheduling with topology awareness - **Oversubscription:** Compute and memory oversubscription for higher density - **Multi-vendor Support:** NVIDIA, Ascend, Cambricon, Hygon DCU, Iluvatar - **Intelligent Operations:** Full observability, monitoring, and alerting **Enterprise Services:** Dedicated technical partnership, 24/7 mission-critical support, GPU optimization consulting, continuous platform evolution, training programs, ecosystem integration. **Pages:** [Chinese](https://www.theriseunion.com/zh/product/vast.html) | [English](https://www.theriseunion.com/en/product/vast.html) --- ### Rise CAMP — AI Workload Orchestration Platform High-performance AI workload orchestration built on Rise VAST with intelligent scheduling and automated resource allocation. **Key Metrics:** 10x GPU utilization | 70% cost reduction | 5-minute deployment | 99% hardware compatibility **Features:** - **Multi-element Unified Scheduling:** GPU, CPU, memory, storage orchestrated together - **GPU Virtualization:** Fine-grained fractional GPU allocation - **Training & Fine-tuning:** Distributed training and model fine-tuning with intelligent resource allocation - **Inference Acceleration:** Optimized for small model and inference workloads - **Security & Compliance:** Multi-tenant isolation, audit logging, RBAC - **Intelligent Operations:** Automated monitoring, alerting, and capacity planning **Pages:** [Chinese](https://www.theriseunion.com/zh/product/camp.html) | [English](https://www.theriseunion.com/en/product/camp.html) --- ### Rise MAX — AI Compute Appliance All-in-one hardware+software appliance integrating GPU virtualization (VAST), intelligent scheduling (CAMP), and unified management. **Key Metrics:** 1-hour deployment | 50% cost reduction | Second-level elastic scaling | 99% hardware compatibility **Features:** - **Out-of-the-Box:** Pre-configured, production-ready in 1 hour - **Centralized Management:** Single pane of glass for all compute resources - **Elastic Scaling:** Scale GPU resources up/down in seconds - **Open Ecosystem:** Compatible with mainstream AI frameworks and tools **Benefits:** Cost efficiency through VAST/CAMP optimization, performance assurance with dedicated resources, rapid deployment, enterprise security and reliability. **DeepSeek Solution (MAX-DS):** Pre-loaded DeepSeek model series from 1.5B to 671B parameters, ready for enterprise AI inference. **Pages:** [Chinese](https://www.theriseunion.com/zh/product/max.html) | [English](https://www.theriseunion.com/en/product/max.html) --- ### Rise ModelX — AI Model Management Platform (MaaS) Next-generation enterprise Model-as-a-Service platform with integrated model repository, AI Gateway, training/inference pipeline, and Agent orchestration. **Features:** - **Model Repository:** Centralized model lifecycle management - **Training/Inference Integration:** Unified pipeline from training to production serving - **Heterogeneous Compute Adaptation:** Automatic model deployment across GPU types - **Auto-Scaling:** Dynamic resource allocation based on inference load - **Agent/MCP Support:** Built-in Agent orchestration and MCP protocol support **Enterprise AI Gateway (ensuring model service stability):** - OpenAI-compatible unified API - High availability with circuit breaking, failover, and load balancing — guarantees stable model serving in production - Security: authentication, authorization, rate limiting - Intelligent routing: route requests across model replicas and heterogeneous backends - Traffic control with canary deployment and graceful degradation - Full observability with metrics, logs, and traces **Use Cases:** Enterprise knowledge Q&A, high-concurrency inference serving, vertical industry model customization, AI agent development. **Pages:** [Chinese](https://www.theriseunion.com/zh/product/modelx.html) | [English](https://www.theriseunion.com/en/product/modelx.html) --- ## Technology Stack ### HAMi — Heterogeneous AI Computing Virtualization Middleware Open-source CNCF sandbox project maintained by RiseUnion. - **Repository:** https://github.com/Project-HAMi/HAMi - **Purpose:** GPU sharing, scheduling, and resource management in Kubernetes - **Supported Hardware:** NVIDIA GPU, Huawei Ascend, Cambricon MLU, Hygon DCU, Iluvatar CoreX - **Adoption:** 1000+ enterprise users worldwide - **Capabilities:** vGPU slicing, memory isolation, compute quota enforcement, topology-aware scheduling ### Product Architecture (Bottom-Up) 1. **Foundation Layer:** Physical GPUs (NVIDIA, Ascend, Cambricon, Hygon, Iluvatar) + Kubernetes 2. **Resource Layer:** Rise VAST — compute pooling, vGPU virtualization, oversubscription 3. **Orchestration Layer:** Rise CAMP — workload scheduling, multi-tenant management, operations 4. **Application Layer:** Rise ModelX — model serving, AI Gateway, Agent orchestration Rise MAX packages layers 1–3 as a turnkey appliance. --- ## Company Information - **Full Name:** Beijing RiseUnion Technology Co., Ltd. (北京睿思智联科技有限公司) - **Focus:** AI compute infrastructure — virtualization, scheduling, and management of heterogeneous GPU resources - **Core Values:** Connect (连接), Empower (赋能), Autonomy (自主) - **Vision:** Making AI computing accessible and efficient for every organization - **Partnership:** Strategic collaboration with 4Paradigm on VAST platform - **Funding:** Backed by Plum Ventures (梅花创投) - **Contact:** contact@riseunion.io | +86 400-605-2336 | [Contact page](https://www.theriseunion.com/en/contact.html) --- ## Blog Topics The company blog covers: - **Technical Guides:** GPU virtualization, HAMi usage, model training (PyTorch, DeepSeek), Kubernetes GPU management - **Product Updates:** New features, DeepSeek appliance, platform releases - **Company News:** Funding, partnerships, industry events, awards - **Industry Analysis:** AI computing trends, heterogeneous computing, cloud-native AI **Blog:** [Chinese](https://www.theriseunion.com/zh/blog.html) | [English](https://www.theriseunion.com/en/blog.html)