Learn AI Engineering
by building.
A hands-on curriculum that teaches RAG, Agentic RAG, multi-agent systems and agent teams through small, self-contained projects. Every demo runs entirely in your browser with mock models and static data — no API keys, nothing to install.
Why this works
🧪 Learn by doing
Each technique is a tiny, runnable project. Tweak the input, run the pipeline, and watch the reasoning trace step by step.
🐍 Python + 🟨 JavaScript
Every project ships both implementations side by side. The JS version runs live here; the Python version is yours to run locally.
📖 Read the mechanics
Technical notes on LLMs, prompting and architecture are woven into every project, so you understand why, not just how.
The projects
Open a project and hit “Mark as complete” to start tracking. Saved on this device only.
01 · Prompt Playground
Discover how a prompt's structure — system persona, user request, and grounding context — shapes a model's output.
02 · Embeddings & Vector Search
Turn text into vectors and rank documents by cosine similarity — the retrieval engine behind every RAG system.
03 · Tool / Function Calling
Let the model decide when to call a tool, run it, and fold the result back into the answer — the loop that creates agents.
04 · Retrieval-Augmented Generation (RAG)
Retrieve relevant context, then generate a grounded answer. The canonical pattern for trustworthy LLM answers.
05 · Agentic RAG
Add a reasoning loop around retrieval: judge relevance, rewrite the query, and retrieve again before answering.
06 · Multi-Agent System
Split a task across specialised agents — researcher, analyst, writer — that hand off to one another.