Get fluent in data, AI, and the math behind them, ten minutes at a time.

The curriculum

8 courses, 186 lessons

• AI Engineering

  7 modules · 22 lessons

  Build reliable products on foundation models: get usable outputs from prompts, ground answers with retrieval, design agents and tools, evaluate with real test sets, decide when to fine-tune, and run LLM features fast, cheaply, and safely in production.

  1. 1Building with foundation models
  2. 2Retrieval-augmented generation
  3. 3Agents and tool use
  4. 4Evaluating LLM applications
  5. 5Adapting models to your task
  6. 6LLMs in production
  7. 7Safety and security

• AI Theory

  8 modules · 24 lessons

  How modern AI models work inside: neural networks and backpropagation, training deep networks, convolutional and recurrent architectures, attention and transformers, language models and scaling laws, reinforcement learning, fine-tuning and alignment, and generative models for images and beyond.

  1. 1Neural network foundations
  2. 2Training deep networks
  3. 3Architectures for structured data
  4. 4Transformers
  5. 5Language models
  6. 6Reinforcement learning
  7. 7Adapting and aligning models
  8. 8Generative models beyond text

• Data Science and Machine Learning Theory

  8 modules · 23 lessons

  Understand why the models you use every day work, and when they fail: generalization and the bias-variance trade-off, linear models and GLMs, gradient-based optimization, trees and boosting, kernels, clustering and dimensionality reduction, cross-validation and calibration, missing data, forecasting, and model interpretation.

  1. 1How learning works
  2. 2Linear models
  3. 3Optimization for learning
  4. 4Trees and ensembles
  5. 5Distances, margins, and kernels
  6. 6Unsupervised learning
  7. 7Model selection and calibration
  8. 8Theory for everyday data science

• ML System Design

  7 modules · 20 lessons

  Design machine learning systems that survive production: frame the problem, build the data and feature pipelines, choose and evaluate models, serve them under latency budgets, and keep them healthy once real users arrive. Finish with three end-to-end case studies.

  1. 1Foundations
  2. 2Data
  3. 3Features
  4. 4Modeling and Evaluation
  5. 5Deployment and Serving
  6. 6Operating ML in Production
  7. 7Case Studies

• Machine Learning Engineering

  7 modules · 22 lessons

  Build, ship, and run ML code like production software: structure and test ML projects, validate and version data, make training fast on GPUs, package and serve models, speed up inference, automate the path to production, and keep the bill under control.

  1. 1From notebook to production code
  2. 2Data pipelines for ML
  3. 3Training infrastructure
  4. 4Packaging and serving models
  5. 5Inference performance
  6. 6Automation and operations
  7. 7Cost and platforms

• Mathematical Methods for Physicists and Engineers

  9 modules · 27 lessons

  The analytical toolkit of physics and engineering, problem first: series and complex numbers, linear algebra, vector calculus, differential equations, Fourier and Laplace transforms, complex analysis, PDEs and Green's functions, special functions, variational methods, and tensors.

  1. 1Series and complex numbers
  2. 2Linear algebra
  3. 3Vector calculus
  4. 4Ordinary differential equations
  5. 5Fourier and Laplace transforms
  6. 6Complex analysis
  7. 7Partial differential equations
  8. 8Special functions
  9. 9Variational methods and tensors

• Numerical and Computational Methods for Physicists and Engineers

  9 modules · 27 lessons

  Compute what you cannot solve by hand, and know when to trust the answer: floating point and error, numerical linear algebra, root finding and optimization, interpolation and integration, ODE and PDE solvers, the FFT, Monte Carlo methods, and fast parallel code.

  1. 1Numbers and error
  2. 2Numerical linear algebra
  3. 3Nonlinear equations and optimization
  4. 4Interpolation, differentiation, and integration
  5. 5ODE solvers
  6. 6PDE methods
  7. 7Fourier methods
  8. 8Monte Carlo methods
  9. 9High-performance computing

• Probability Theory and Statistics

  7 modules · 21 lessons

  The probability and statistics that data and ML work runs on: reason about uncertainty, choose the right distribution, estimate with honest error bars, test hypotheses without fooling yourself, analyze experiments, and put confidence intervals on model comparisons.

  1. 1Probability foundations
  2. 2Distributions
  3. 3Sampling and limit theorems
  4. 4Estimation
  5. 5Hypothesis testing
  6. 6Experiment statistics
  7. 7Statistics for machine learning