ML Forge is a visual desktop application that provides a three-tab workflow (Data Prep, Model, Training) for building computer vision pipelines. Users drag nodes onto a canvas, connect them with wires, and the tool automatically generates PyTorch training code. It supports popular datasets (MNIST, CIFAR10, CIFAR100, FashionMNIST, ImageFolder) and provides live training visualization with loss curves, checkpoint saving, and model inference. The tool exports projects into clean, runnable PyTorch code.