Polyhedrontech Company Limited is a Hong Kong-based AI infrastructure startup building the runtime layer for edge and on-device AI. Our mission is to make large language models and multimodal AI faster, more private and more cost-efficient to run across heterogeneous hardware including CPUs, GPUs and NPUs.
Our technology combines model optimisation, quantisation, memory optimisation, heterogeneous computing, automatic hardware detection, model conversion and intelligent resource scheduling into a lightweight inference engine and runtime. Through OpenAI-compatible APIs and integration with modern AI application frameworks, Polyhedrontech enables developers and enterprises to deploy AI applications across PCs, mobile devices, robotics and other edge environments with lower latency, stronger data privacy and reduced cloud dependency.
Areas of activity
Manufacturing
Manufacturing
Public Utilities
ICT And Data Center
Technology
Building And Construction TechnologyEducation TechnologyFin-TechHealth TechnologyIntelligent ManufacturingMedical TechnologySupply Chain TechnologySmart City TechnologyProperty TechnologyIoT
Urban Development
Smart City DevelopmentTechnology ParkFree Trade Zone
We have innovated research and experience by deploy and optimising LLMs and multimodal AI on edge and local devices with model optimisation, quantisation, heterogeneous CPU/GPU/NPU acceleration, runtime integration and private AI deployment.
Hands-on training program for robotics secondary development, covering robot SDK/API development, AI and LLM integration, computer vision, voice interaction, edge AI deployment and practical robot application development.