DeGirum’s hardware-plus-software platform accelerates the development and deployment of AI applications at the edge. The company’s first-generation hardware accelerator, ORCA, is a flexible and powerful compute engine with up to 4 TOPS of inference capacity. ORCA’s capability to efficiently process pruned networks provides a significant performance boost; combined with dedicated DRAM, it enables efficient multiplexing of multiple models without sacrificing accuracy and performance. DeGirum’s Universal Inference Runtime software, DgRT, abstracts model inferencing to a function call agnostic to the hardware, model, connectivity, and operating system. DgRT enables application developers to target diverse edge hardware using the same software stack. DeGirum’s Model Designer, DgGraph, is a GUI tool for designing and visualizing machine learning models. DgGraph delivers model portability across varying hardware, ensuring that the model definition is compatible with downstream tools for compilation and optimization.