A VM/Containerized Approach for Scaling TinyML Applications