36+ interactive visualizations

See algorithms
execute.

Step through sorting algorithms, neural networks, and transformer architectures — one operation at a time, with real code.

bubble-sort.js
function bubbleSort(arr) {
for (let i = 0; i < arr.length; i++) {
if (arr[j] > arr[j + 1]) {
swap(arr, j, j + 1);
}
}
}
36+
Interactive Lessons
13
Sorting Algorithms
15
Blog Articles
100+
Step-by-Step Visualizations

Pick a track

Each track is a structured learning path with interactive visualizations, step-by-step explanations, and quizzes.

What you'll learn

Sorting Algorithms

Master 13 sorting algorithms through interactive visualizations. Start with simple O(n²) algorithms like bubble sort, insertion sort, and selection sort — watching elements compare and swap in real time. Progress to efficient O(n log n) algorithms: quick sort with pivot partitioning, merge sort with divide-and-conquer merging, heap sort with binary heap operations, and shell sort with gap-based comparisons. Explore non-comparison sorts like radix sort, counting sort, and bucket sort that achieve linear time. See Tim Sort — the hybrid algorithm used by Python and Java's standard libraries. Every algorithm includes synchronized code highlighting, performance statistics, and complexity analysis.

Neural Networks & Deep Learning

Build understanding of neural networks from first principles. Begin with the perceptron learning rule and decision boundaries. Follow data through a forward pass — watching weights multiply inputs, activation functions introduce non-linearity, and layers combine features into predictions. Understand loss functions (MSE, cross-entropy) that measure prediction errors. Watch gradient descent navigate loss landscapes, and trace backpropagation as gradients flow backward through layers via the chain rule. Explore CNNs with convolution filters detecting edges and features, RNNs with hidden states processing sequences, and embeddings mapping words to meaningful vectors.

Transformer Architecture

Dissect the architecture behind GPT, BERT, and modern AI in 10 focused lessons. Start with self-attention — the mechanism that lets every token directly interact with every other token. Learn Query, Key, and Value projections, see attention scores computed as dot products, and watch softmax normalize them into probability distributions. Explore multi-head attention with side-by-side heatmaps showing different heads capturing different relationships. Understand positional encoding, feed-forward networks, residual connections, and layer normalization. Then compare encoder (BERT), decoder (GPT), and encoder-decoder (T5) architectures.

Large Language Models

Understand how ChatGPT, Claude, and modern LLMs work from tokenization to deployment. See how BPE tokenization splits text into subwords, how pre-training learns language patterns from billions of tokens via next-token prediction, and how scaling laws predict performance from model size. Follow the text generation process: probability distributions over vocabulary, sampling strategies (temperature, top-k, top-p) that control creativity vs coherence. Learn how fine-tuning adapts models to specific tasks, how RLHF aligns models with human preferences, and how prompt engineering techniques (zero-shot, few-shot, chain-of-thought) unlock model capabilities.

From the blog

In-depth articles on algorithms, machine learning, and computer science — each linking to interactive visualizations. Topics include bubble sort, quick sort, merge sort, binary search, Big O notation, neural networks, backpropagation, CNNs, LSTMs, Word2Vec, activation functions, and overfitting.

Read articles