This chapter provides a detailed overview of machine learning concepts essential for scientific and engineering applications, with particular emphasis on automatic differentiation, deep learning architectures, and computational implementation frameworks. We begin with an overview of automatic differentiation (AD), contrasting forward and reverse modes while explaining their computational efficiency trade-offs, implementation using dual numbers and computational graphs, and applications in gradient-based optimization. Next, we survey fundamental deep learning architectures including feedforward neural networks, convolutional neural networks (CNNs), residual neural networks, and recurrent neural networks (RNNs), along with key concepts such as activation functions, regularization techniques, optimization algorithms, and hyperparameter tuning strategies. We also discuss advanced topics including neural architecture search, emerging frameworks like physics-informed neural networks and neural operators (DeepONet and Fourier Neural Operators), and the integration of generative AI models in computational mechanics. Finally, we present a systematic framework for implementing scientific machine learning solutions, outlining the essential components including loss function formulation, neural network architecture selection, PDE integration, domain discretization, solver configuration, numerical integration methods, and utility modules for automatic differentiation and transfer learning.