NewInvite a friend: you both unlock a bonus lesson

Learn AI engineering by doing it.

Bite-sized, hands-on lessons on how real AI products get built: LLMs, prompting, RAG, agents, evals and production. Write real Python in your browser. You can't move on until you've actually got it.

Start learning free See the curriculum

Your first module (5 lessons) free, no card 36 lessons · 212 steps 43 Python exercises

RAG · Lesson 1

Real Python, in your browser

Implement cosine similarity: the math behind every vector search.

▍

▶ Check answer 4/4 tests passed

Correct! First try. +10 XP

Tokens ✦Temperature ✦Prompt engineering ✦Structured output ✦Tool calling ✦Embeddings ✦Chunking ✦RAG ✦Hybrid search ✦Reranking ✦Agents ✦MCP ✦Evals ✦LLM-as-judge ✦Prompt injection ✦Caching ✦Gradient descent ✦Fine-tuning ✦Tokens ✦Temperature ✦Prompt engineering ✦Structured output ✦Tool calling ✦Embeddings ✦Chunking ✦RAG ✦Hybrid search ✦Reranking ✦Agents ✦MCP ✦Evals ✦LLM-as-judge ✦Prompt injection ✦Caching ✦Gradient descent ✦Fine-tuning ✦

How it works

No 3-hour videos. Just practice.

Each lesson is 5–10 minutes. Every step asks you to do something, and you get instant feedback.

Read one idea

A short, precise explanation of one concept, with animated diagrams of how data flows through real systems.

Do it yourself

Write Python, drag pipeline steps into order, tune live widgets, sort and match. You must get it right to continue.

Get instant feedback

Hints when you're stuck, explanations when you're right, XP and levels to keep you moving.

Every lesson ends in something real, like this RAG pipeline:

User question

Embed

Hybrid search

Rerank

Prompt + sources

LLM

Cited answer

Know the difference

AI engineer ≠ ML engineer ≠ software engineer.

They overlap, but the daily work is very different. This course trains AI engineers, and teaches enough ML to work well with ML engineers.

Software Engineer

Builds reliable systems with deterministic code.

APIs, databases, infrastructure
Testing & CI/CD
Output: apps & services

ML Engineer

Trains, tunes and serves models from data.

Data pipelines & features
Training, tuning, MLOps
Output: trained models

This course

AI Engineer

Builds products on top of foundation models.

Prompts, context & RAG
Tools, agents & evals
Output: LLM-powered features

Full comparison + "which path fits me?" quiz

Curriculum

From "what's a token?" to shipping agents.

Module 15 lessons

Foundations

What AI engineering is, and how LLMs really work under the hood.

Your First Day as an AI Engineer
Tokens: What Models Actually Read
Next-Token Prediction & Temperature
How LLMs Are Built (and Why They Hallucinate)
Context Windows, Latency & Cost

Start module

Module 26 lessons

Prompting & LLM APIs

Talk to models like an engineer: messages, prompts, JSON and tools.

Anatomy of an LLM API Call
Streaming Responses
Prompt Engineering That Works
Reasoning Models & Chain of Thought
Structured Output (JSON)
Tool Use (Function Calling)

Start module

Module 35 lessons

Retrieval (RAG)

Ground models in your own data with embeddings, chunking and search.

Embeddings & Similarity
Chunking Documents
Building a RAG Pipeline
Hybrid Search & Reranking
Evaluating RAG: Recall, MRR & Faithfulness

Start module

Module 45 lessons

Agents

From single calls to systems that plan, use tools and act.

Workflows vs Agents
The Agent Loop
Memory for Assistants & Agents
Multi-Agent Systems
MCP: Model Context Protocol

Start module

Module 55 lessons

Evals & Production

Measure quality, defend against attacks, and ship systems that hold up.

Evals: Testing LLM Systems
Prompt Injection & Guardrails
Reliability, Cost & Observability
Observability: Traces, Latency & Cost
Cutting LLM Costs: Caching, Batching & Routing

Start module

Module 64 lessons

ML for AI Engineers

The machine-learning core you need to work with ML engineers, and to know when to fine-tune.

Neural Networks in One Lesson
How Models Learn: Gradient Descent
Data Splits, Overfitting & Leakage
Prompting vs RAG vs Fine-Tuning

Start module

Module 76 lessons

Build an AI SaaS

From idea to paying users: economics, safe APIs, privacy, limits, UX and launch.

Pick an AI Feature People Pay For
Call AI From Your App, Safely
Protect User Data
Usage Limits & Pricing That Cover AI Costs
AI UX People Trust
Launch, Measure, Improve

Start module

Made for curious minds who learn by doing.

Try before you pay

Your first module (5 lessons) is free, with no card and no sign-up. Progress is saved as you go.

5–10 minute lessons

Fits a lunch break. Resume exactly where you stopped.

Real Python

Your code runs in real CPython (Pyodide) right in the page and is checked by hidden tests.

Accurate, current content

Based on how teams actually build: published research, OWASP and provider guidance.

Keyboard-friendly

Enter to continue, ⌘/Ctrl+Enter to check code, Tab to indent.

Auto-advance

Get it right and the next step slides in. Pause any time.

Free guides

Start learning AI today, free.

In-depth, plain-English guides to the most-asked questions about AI, each linked to hands-on practice.

All guides

9 min read

Questions? Answers.

What is keepUp AI? +

keepUp AI is an interactive course that teaches AI engineering by doing: short explanations followed by hands-on tasks, real Python exercises that run in your browser, interactive labs, quizzes and real-world case studies.

Is it good for beginners? +

Yes. If you can read basic Python you can start today. Every AI idea, from tokens to agents, is taught from zero with plenty of practice.

Can I try it for free? +

Yes. The whole first module (5 lessons) is completely free, including Keepy the AI tutor, with no card and no sign-up. You can also unlock bonus lessons by inviting friends.

What will I be able to build? +

LLM features with validated outputs, RAG assistants over your own documents, tool-using agents, MCP servers and eval suites. The final module takes you from idea to a paying AI SaaS: unit economics, a secure backend, privacy, usage limits, AI UX and launch metrics. Plus 36 copy-paste templates to start from.

How is this different from video courses? +

You can't move on until you've actually done the thing. Every step asks you to write code, tune a lab, solve a quiz or analyze a real case, so the skills stick.

Launch offer

Pay once. Keep it forever.

Lifetime access is $69 instead of $139 while we launch, with every future lesson included. Prefer a subscription? Yearly works out to just $4.08/month.

See plans Try your first module (5 lessons) free

Best value

Lifetime

$139$69one-time

Everything, forever: no renewals
Unlimited Keepy AI tutor, forever
Every future lesson, module & template
Price locked: never pay again
Costs less than 8 months of Monthly

14-day money-back guarantee Secure checkout by Polar Instant access

Your first lesson takes 6 minutes.

Find out what AI engineers actually do, and how it differs from ML and software engineering.

Start lesson 1

Learn AI engineering by doing it.

No 3-hour videos. Just practice.

Read one idea

Do it yourself

Get instant feedback

AI engineer ≠ ML engineer ≠ software engineer.

Software Engineer

ML Engineer

AI Engineer

From "what's a token?" to shipping agents.

Foundations

Prompting & LLM APIs

Retrieval (RAG)

Agents

Evals & Production

ML for AI Engineers

Build an AI SaaS

Made for curious minds who learn by doing.

Try before you pay

5–10 minute lessons

Real Python

Accurate, current content

Keyboard-friendly

Auto-advance

Start learning AI today, free.

How to Learn AI Engineering in 2026: A Practical Roadmap

What Is an AI Engineer? Role, Skills and How to Become One

What Is RAG (Retrieval-Augmented Generation)? A Clear Guide

Prompt Engineering Guide: Techniques That Work in Production

What Are AI Agents? How They Work and When to Use Them

What Is MCP (Model Context Protocol)? A Beginner's Guide

LLM Evals: How to Test and Measure AI Applications

Learn AI for Beginners: Where to Start (Free, Hands-On)

Handy calculators for AI builders

AI words, explained simply

Questions? Answers.

Pay once. Keep it forever.

Your first lesson takes 6 minutes.