Skip to main content
About Us

RiseUnion Technology Co., Ltd.

AI heterogeneous compute management experts · HAMi open source core maintainers

RiseUnion Technology Co., Ltd. is an innovative technology company focused on AI compute management and scheduling, building industry-leading data center-grade AI compute management platforms. Through a self-developed architecture and open ecosystem, we enable deep integration of domestic chips with AI applications, solving heterogeneous GPU resource management, compute pooling, and task scheduling challenges.
HAMi (Heterogeneous AI Computing Virtualization Middleware) is a CNCF Sandbox open-source project for GPU sharing, slicing, and hard-isolation scheduling. Rise VAST is the enterprise edition built on HAMi, and RiseUnion is a core maintainer of the HAMi community.
RiseUnion has completed tens of millions RMB in funding (exclusively by Plum Ventures), with strategic investment from 4Paradigm. We are a National High-tech Enterprise and Beijing SRDI SME, certified with ISO 9001, ISO 27001, ISO 20000, and CMMI3, holding 10+ software copyrights and patents. As chair of the AIIC Compute Pooling Working Group, we led the "Heterogeneous Compute Virtualization and Pooling" group standard, co-authored the Diverse Computing Optimization Action Plan, and participated in the National Cloud Computing Standards Conference and Compute & AI Fusion Symposium. Our products serve BYD, Huatai Futures, the National Supercomputing Center in Jinan, Chongqing 2nd Hospital, CEPREI (Saibao Lab) and more flagship customers, with 10+ domestic chip certifications (Ascend, Cambricon, Hygon, KunlunXin and more), with large-scale production deployments across finance, government and defense, transportation, education, and healthcare.

Our Vision

Connect

Connecting isolated compute silos to build a unified, standard compute foundation, making compute as accessible as a utility.

Empower

Empowering enterprise AI transformation with extreme resource efficiency and unified training/inference, lowering adoption costs and accelerating industry intelligence.

Autonomy

Persisting in core technology self-reliance, deeply adapting domestic AI chips to foster a thriving ecosystem and build the foundation for the digital economy.

Milestones

2022

Founded

Beijing RiseUnion Technology Co., Ltd. founded, focused on AI compute management

2024-01

Rise VAST 1.0

Released heterogeneous GPU virtualization platform (HAMi Enterprise Edition). Certified as Innovative SME and Tech-based SME

2024-06

Rise CAMP 1.0

Released compute scheduling platform with 4-way intelligent scheduling

2024-10

AIIC Working Group

Elected chair of AIIC AI Compute Pooling Working Group (guided by MIIT)

2025-01

Enterprise Adoption

Benchmark partnerships with BYD, Huatai Futures, Jinan Supercomputing. National High-tech Enterprise certification

2025-10

Rise Edge 1.0

Released edge compute management platform with cloud-edge coordination

2025-12

Industry Standard

Published group standard "Heterogeneous Compute Virtualization and Pooling System Requirements"

Qualifications & Certifications

Management System Certifications

ISO 9001ISO 27001ISO 20000CMMI3

Intellectual Property

10+
Software Copyrights
10+
Patents

Enterprise Designations

National High-tech EnterpriseBeijing SRDI SMEInnovative SMETech-based SME

Chip Compatibility

Compatibility certified with 10+ domestic chip vendors: Ascend, Cambricon, KunlunXin, Hygon, Iluvatar, Moore Threads, MetaX, Enflame and more

Three-layer Architecture, Full-stack Delivery

VAST

Unified Collection · Full Observability

Rise VAST resource management layer: unified management of 10+ chip vendors, full-stack observability from GPU to model, with auto fault isolation and enterprise alert closed-loop.

CAMP

Fine-grained Slicing · Precise Scheduling

Rise CAMP scheduling layer: vGPU fine-grained slicing with compute overcommit, 4-way aware scheduling (topology/priority/load/resource), domestic chip dynamic partitioning beyond vendor limits.

ModelX

AI as a Service · Clear Operations

Rise ModelX model service layer: unified training & inference AI platform with built-in AI Gateway, model hub, multi-engine inference, and dual-dimension resource/Token metering.