Deep Learning & Modern AI
Neural networks, transformers and generative models.
Deep Networks & Optimization
Master the engineering that makes deep networks trainable: why depth wins, why gradients die or explode, and the precise algorithms (He init, BatchNorm, Dropout, Adam) that solved each problem.
Convolutional Neural Networks
Convolutional Neural Networks exploit the spatial structure of images through local receptive fields, weight sharing, and hierarchical feature extraction — turning raw pixels into rich representations.
Sequence Models & RNNs
Recurrent neural networks model sequences by threading hidden state through time — and every modern language model traces its lineage through the LSTM's elegant solution to vanishing gradients.
Attention & Transformers
Attention is all you need: how a simple query-key-value lookup revolutionized sequence modeling and became the backbone of every frontier AI system.
Large Language Models
From raw text to world-knowledge engines: how GPT-scale transformers are pretrained, scaled, and aligned into the language models that are reshaping computing.
Generative Models
Generative models teach machines to imagine — from encoding latent structure in VAEs to adversarial games in GANs to denoising diffusion that sculpts images from pure noise.
Reinforcement Learning
Reinforcement learning is the mathematics of learning by doing — an agent discovers optimal behavior through trial, error, and reward signals, from Atari games to Go to robotic locomotion.
Embeddings & Representation Learning
Embeddings transform raw tokens into geometry where meaning becomes distance — the foundational trick behind every modern AI system.