Demos
Small AI-built apps. Each one showcases a different AI skill or knowledge.
Visual Learning
Tokenizer Visualizer
Type anything and watch it shatter into the colored chunks a real LLM actually sees — then understand why models can't count the r's in "strawberry."
Explaining a technical concept through an interactive tool, not a wall of text.
Markov Chain Text Generator
Paste any text, train an n-gram model in milliseconds, and watch it generate gibberish-that-almost-makes-sense — the honest ancestor of every LLM.
Making an abstract mechanism (next-token prediction) tangible through a hands-on toy.
Tiny Neural Net Playground
Place dots on a canvas, press Train, and watch a small neural network literally learn the boundary between them — weights, loss curve and all.
Making an abstract mechanism (gradient descent) tangible through direct manipulation, not just a diagram.
Applied Apps
WebLLM Persona Chat
A real LLM downloads into your browser and runs on your GPU — chat with a jargon-armored Tech Lead, an overhyped Investor, or a User who doesn't know what they want, then read the exact system prompt causing it.
Showing that persona and behavior are mostly just the system prompt, with zero backend and zero API keys.
LinkedIn HDR Brightener
Re-encodes an image so its highlights render brighter than the white of the page around them — a clone of a tool that sells this as magic, with the colour maths and the costs both on show.
Reading a competitor's trick off its own bundle and rebuilding it in an afternoon — then being honest about what it costs.
Games
GPU Rush
A cute factory-idle game: route glowing tokens down conveyor belts into hungry GPUs, keep the context window from overflowing, and upgrade your rig from a potato laptop to a datacenter aisle.
Shipping pure fun — retention mechanics and juice, not just explainers.
Teach the Robot Chinese
Adopt Xiǎodòu (小豆), a cute robot student, and teach it Chinese characters flashcard-by-flashcard. It genuinely learns — a tiny real learning model under the hood — and makes adorably plausible mistakes.
Machine learning as a pet — an honest, inspectable model wearing a mascot's face.