1. Responsible for SDK development in the direction of computer vision and algorithm deployment and performance optimization on the edge NPU;
2. Follow up on model quantization and optimization processes, and modify models based on hardware optimization;
3. Design a model evaluation system based on business scenarios, and improve accuracy through data augmentation, loss function optimization, and other means.
1. Bachelor's degree or above, strong C++/Python development and algorithm engineering capabilities, as well as paper reading and reproduction capabilities;
2. Familiar with model compression algorithms such as quantization, possess good communication and collaboration skills, and are passionate about technical challenges and business implementation;
3. Mobile application development or Unity experience is preferred.
4. Intern for more than 4 days a week, lasting more than 6 months, and those who can start as soon as possible are preferred.