This Week 2 first session of our AI/ML Bootcamp will bridge the gap between traditional machine learning and deep learning by first introducing fundamental algorithmic thinking and key ML concepts such as decision trees, k-nearest neighbors, linear regression, SVMs, overfitting and underfitting, evaluation metrics, and confusion matrices to build a proper foundation. From there, we will transition into the logic of artificial neural networks, explaining how neurons operate through weighted inputs, summation, bias, and activation functions, and discussing why non-linearity matters with examples like ReLU and tanh. Participants will develop a clear understanding of layer structures, forward and backward propagation, optimization techniques such as SGD and Adam, and essential training terminology including epochs, loss, and regularization. We will construct a simple neural network step-by-step in Keras to solidify core ideas before progressing to convolutional neural networks, clarifying why visual data requires specialized architectures and how convolution, pooling, and fully connected layers allow models to automatically detect patterns such as edges, textures, and shapes. Throughout the session, foundational mathematical and conceptual ideas - including matrices, tensors, and the flow of data through a network - will be explained in an accessible and intuitive way, ensuring all participants, regardless of background, can comfortably follow the progression from classic ML methods to modern deep learning techniques while gaining hands-on practice in Keras and TensorFlow.