Gwang-jin Cho

Nota AI develops neural network compression and optimisation technology for deploying AI models on edge devices and embedded systems with limited compute and power resources, providing tools that compress, quantise, and prune large neural network models into lightweight versions suitable for smartphones, surveillance cameras, automotive chips, and IoT devices. The Seoul company NetsPresso platform automates the model optimisation workflow, reducing the engineering effort required to adapt research models for production hardware deployment.\n\nThe company raised approximately $10 million in venture funding from Korean AI-focused investors. Nota AI targets computer vision and NLP model deployment scenarios where inference must run locally on resource-constrained hardware rather than in the cloud, a requirement driven by latency, privacy, and connectivity considerations. The company customers include semiconductor companies integrating AI features into SoC designs and device manufacturers seeking to add on-device AI capabilities without redesigning around more powerful processors.\n\nNota AI competes in the edge AI model optimisation market against Kneron, Hailo, and embedded AI optimisation tools from semiconductor companies including ARM, Qualcomm, and MediaTek. Its software-first approach, which optimises models for deployment across diverse existing hardware rather than requiring a specific custom chip, addresses the fragmented edge hardware ecosystem where AI model developers need to support many different target processors from different vendors. The Korean semiconductor industry concentration creates a proximate customer base among chipmakers and device manufacturers seeking edge AI capabilities.