Topics

Mathematical Foundations

Linear Algebra

Vectors, matrices, projections, and transformations that underlie modern machine learning representations.

Mathematical Foundations

Probability

Random variables, distributions, uncertainty, and likelihood for learning from data.

Mathematical Foundations

Optimization

Objective functions, gradients, convexity, and numerical methods for fitting models.

Classical Machine Learning

Gradient Descent

Iterative parameter updates that move models toward lower training loss.

Classical Machine Learning

Supervised Learning

Learning predictive functions from labeled examples and measured errors.

Classical Machine Learning

Feature Engineering

Transforming raw signals into useful model inputs and inductive biases.

Deep Learning

Neural Networks

Layered differentiable models built from parameters, activations, and composition.

Deep Learning

Backpropagation

Efficiently computing gradients through networks using the chain rule.

Natural Language Processing

Embeddings

Dense vector representations that place related items near one another in learned space.

Natural Language Processing

Tokenization

Converting text into model-readable tokens while balancing vocabulary and sequence length.

Deep Learning

Attention

A differentiable mechanism for routing information between positions according to learned relevance.

Large Language Models

Transformers

Sequence models built from attention, residual streams, normalization, and feed-forward blocks.

Large Language Models

Pretraining

Large-scale self-supervised learning that creates broadly capable foundation models.

Large Language Models

Fine-tuning

Adapting pretrained models to specific tasks, styles, or domains with additional training.

Large Language Models

Prompting

Designing instructions, examples, and context to steer model behavior at inference time.

AI Systems and Evaluation

RAG

Retrieval-augmented generation grounds model responses in external documents and search results.

Large Language Models

RLHF

Preference optimization that aligns model outputs using human feedback and reward models.

AI Systems and Evaluation

Evaluation

Measuring quality, robustness, safety, and usefulness with offline and human-centered tests.

AI Systems and Evaluation

Quantization

Compressing model weights and activations to reduce memory and inference cost.

AI Systems and Evaluation

AI Agents

Systems that plan, use tools, retrieve context, and iterate toward goals with model calls.