Learn
Learn how AI really works
Everything here runs live in your browser: guided pathways, ~28 interactive demos of LLMs, machine learning, responsible AI and the maths under the hood, plus a curated video shelf.
Guided learning pathways
Ordered sequences of demos and videos with progress saved in your browser — no account needed. Pick one and work through it at your own pace.
How an LLM works, end to end
17 stepsThe flagship journey — from raw text to a grounded answer. Interactive demos interleaved with the best explainer videos, in learning order.
See pathway →AI for the curious
9 stepsNo maths degree required. A gentle, visual path from “what is a neural network?” to how modern AI writes and reasons.
See pathway →For decision-makers
8 stepsThe concepts behind responsible enterprise AI — what LLMs and RAG are, where bias and fraud trade-offs bite, and how to size up your own adoption.
See pathway →Responsible & safe AI
7 stepsbigspark's specialism, made hands-on. Why safety matters, then the real enterprise trade-offs — prompt-injection guardrails, bias and fairness, fraud thresholds — and the alignment ideas behind them.
See pathway →Classic ML foundations
8 stepsBefore the transformers — the workhorse algorithms every data scientist knows, hands-on: regression, classification, clustering and learning by gradient descent.
See pathway →Interactive demos
Every demo runs live in your browser — real maths, no libraries, nothing to install.
Large language models
How modern LLMs actually work — from tokens and attention to sampling and retrieval.
Token Visualiser
Paste any text and watch how a language model chops it into tokens — the subword chunks LLMs actually read.
Open demo →Embeddings Explained
See how AI turns words into vectors — explore a semantic map, find nearest neighbours, and watch analogy arithmetic.
Open demo →Temperature & Sampling
Explore how temperature, top-k and top-p control the randomness of language models.
Open demo →Attention Visualiser
See how transformer attention works — type a sentence and watch each token decide what to attend to.
Open demo →Positional Encodings
How do transformers know word order? Explore the sinusoidal positional-encoding heatmap.
Open demo →Beam Search vs Greedy
See how a model decodes text — compare greedy decoding against beam search.
Open demo →Context Windows & Chunking
Why long documents get sliced up for RAG — set a chunk size and overlap, and watch the split.
Open demo →How RAG Works
Watch Retrieval-Augmented Generation in action — retrieve, augment, generate.
Open demo →Markov Chain Text
Generate text the pre-LLM way — build a word-level Markov model and sample next words.
Open demo →Machine learning
Classic algorithms you can poke at — training, clustering and classification, live in your browser.
Neural Net Trainer
Place two classes of points, hit train, and watch a neural network learn the boundary.
Open demo →k-means Clustering
Watch unsupervised learning find structure — drop points, pick k, step through Lloyd's algorithm.
Open demo →Linear Regression
Drag points and watch the least-squares best-fit line update live, with residuals and R².
Open demo →k-Nearest Neighbours
Classify by proximity — place labelled points and see the decision boundary.
Open demo →Decision Trees & Gini
How trees split data — slide a threshold and watch Gini impurity change.
Open demo →Convolution Kernels
The building block of CNNs — apply edge-detect, blur, sharpen kernels to an image.
Open demo →Q-learning Grid World
Watch an agent learn by trial and error — train tabular Q-learning in a grid world.
Open demo →Responsible AI
Making AI safe, fair and accountable — from bias checks to governance.
Prompt Injection & Guardrails
See how a system prompt, untrusted content and a user question stack into one context.
Open demo →Bias & Fairness
One model, two groups — adjust thresholds and see selection rates and the four-fifths rule.
Open demo →Fraud Threshold Trade-off
Move a fraud model's decision threshold and watch precision, recall, F1 shift in real time.
Open demo →LLM Evaluation & Hallucination
Score candidate answers on token-F1, check grounding against the source, and flag hallucinations.
Open demo →Maths & probability
The maths under the hood, made interactive.
Bayes & Base Rates
Why a 90%-accurate test can still be wrong — set prevalence and see why false positives can dwarf true ones.
Open demo →Monte Carlo π
Estimate π by throwing random darts at a circle — a hands-on look at Monte Carlo sampling.
Open demo →Gradient Descent
Watch optimisation roll downhill across a 2-D loss surface — tune the learning rate.
Open demo →Data engineering
Great AI starts with clean, well-run data — validation, pipelines and observability.
Data Quality Checks
Toggle validation rules over sample records and see what they catch — each highlighted cell is a problem found before it reaches production.
Open demo →Feature Engineering
Models only read numbers. Watch a raw table transform into model-ready features — one-hot encode categories, ordinal encode sizes, and scale to 0–1.
Open demo →For your business
Quick, interactive tools to size up what AI could do for your organisation.
Videos
A curated shelf of talks and tutorials on AI, data engineering and building software well.
Transformers, the tech behind LLMs
A visual walkthrough of the transformer architecture behind modern AI.
Intro to Large Language Models
A one-hour, non-hype primer on how LLMs work and where they're going.
What is Retrieval-Augmented Generation (RAG)?
How RAG grounds LLMs in your own data — the backbone of most enterprise AI.
Andrew Ng on AI Agentic Workflows
Why agentic workflows are driving the next wave of AI progress.
But what is a neural network?
The clearest visual introduction to what a neural network actually is.
Intro to AI Safety, Remastered
A clear, grounded introduction to why AI safety matters.