Minitensor is a lightweight, high-performance tensor operations and automatic differentiation library designed for developers seeking a PyTorch-like experience with enhanced speed and efficiency. Built with a Rust backend, it ensures maximum speed and memory efficiency, while its Python bindings provide a familiar API for easy adoption. Minitensor supports complete neural network layers and optimizers, offering seamless interoperability with NumPy arrays. Key features include automatic differentiation for gradient computation, a modular design for customization, and a high-performance engine optimized with SIMD. The library is currently in active development, making it a cutting-edge choice for developers looking to build and deploy intelligent applications.