AI Engineer Path

Concept Lab

22 interactive simulations for the ideas that are easier to see than to read. Each one also appears inside its lesson.

  1. Week 3 · PrerequisitesHNSW graph searchWatch a query hop down the layers of a navigable small-world graph.
  2. Week 4 · PrerequisitesMatrices move spaceDrag the basis vectors and watch the whole grid transform with them.
  3. Week 4 · PrerequisitesDot product → cosine similarityRotate two vectors and see projection, dot product and cosine update live.
  4. Week 4 · PrerequisitesEigenvectorsFind the directions a matrix only stretches and never turns.
  5. Week 4 · PrerequisitesThe chain rule as a pipelineNudge the input and watch the change multiply through each stage.
  6. Week 4 · PrerequisitesGradient descentRoll a ball down a loss surface. Try tiny and huge learning rates.
  7. Week 5 · PrerequisitesBayes on a population gridSee why a 99%-accurate test for a rare condition is usually wrong.
  8. Week 5 · PrerequisitesCentral Limit TheoremSample from a skewed distribution and watch the means form a bell curve.
  9. Week 5 · PrerequisitesIs prompt B really better?Simulate eval sets of different sizes and see how often noise fools you.
  10. Week 6 · Machine LearningBias–variance tradeoffRaise model complexity and watch train error fall while test error turns back up.
  11. Week 9 · Machine LearningDecision tree splitsWatch a tree carve 2D space into rectangles, one split at a time.
  12. Week 10 · Machine Learningk-means, step by stepAssign, move centroids, repeat, until nothing changes.
  13. Week 14 · Other ModalitiesConvolution kernelsSlide a 3×3 kernel over an image and see edges appear.
  14. Week 15 · Deep LearningBackpropagation, node by nodeRun a forward pass, then watch gradients flow back through a tiny network.
  15. Week 16 · Deep LearningOptimiser raceSGD, Momentum, RMSprop and Adam race across the same awkward valley.
  16. Week 19 · Deep LearningSelf-attention, Q·K·VPick a word and see which other words it attends to, and why.
  17. Week 25 · Build a GPTByte-pair encodingMerge the most frequent pair, again and again, and watch tokens form.
  18. Week 26 · AI Engineer TrackTemperature and top-pReshape a next-token distribution and sample from it.
  19. Week 26 · AI Engineer TrackPrefill, decode and the KV cacheSee why input and output tokens cost differently, and what caching saves.
  20. Week 31 · AI Engineer TrackA RAG pipelineChunk → embed → retrieve → rerank → answer, with each stage visible.
  21. Week 33 · AI Engineer TrackLoRA: low-rank updatesCompare a full weight update with W + BA and count the parameters saved.
  22. Week 34 · AI Engineer TrackThe ReAct agent loopThought → action → observation, including a failing tool and a retry.