Practice AI Engineering
Scrimba's AI catalog covers prompting, RAG, agents, MCP, and deployable AI UIs across 12 courses, all as interactive scrims where you edit the prompts and tool calls yourself instead of just watching. Start with the AI Engineer Path once you can already ship a basic web app; pick a single course below if you only need one skill.
Who this is for
Web developers who can call an API and want hireable AI features, not a second PhD in statistics. AI engineering here means shipping products with models, not notebook-only experiments.
Why practice AI engineering on Scrimba?
- You build real apps: full applications with APIs, databases, and interfaces, not toy notebooks.
- The format is interactive: pause any scrim and change the prompt or model to see what happens.
- 12 AI courses cover RAG, agents, MCP, and more.
- No ML degree required. Know JavaScript and you can start today.
Top AI engineering courses on Scrimba
AI Engineer Path
Duration: 11.4 hrs | Level: Intermediate | Access: Pro
The structured path from prompt engineering to production AI agents, with several hands-on projects.
Learn AI Agents
Level: Intermediate | Duration: 2.0 hrs | Access: Pro
Bob Ziroll builds a ReAct agent by hand over nine scrims (a loop capped at five iterations, regex parsing of the model's Action: lines), then rewrites it on OpenAI function calling. Six challenges, all Pro; the first five scrims are free. Bring your own OpenAI key.
Learn RAG (Retrieval-Augmented Generation)
Level: Intermediate | Duration: 1.6 hrs | Access: Pro
Query your own data: OpenAI embeddings for ten made-up podcasts, a Supabase table with a vector(1536) column, LangChain chunking, and a ReelRecs movie chatbot. Three challenges are AI-checked. The first five scrims (about 21 minutes) are free; the vector database onward is Pro.
Intro to Model Context Protocol (MCP)
Duration: 37 min | Access: Pro
Three build scrims carry the course: a weather MCP server in TypeScript with one tool, a weather://cities resource, and a stdio transport, then MCP Inspector and Claude Desktop as clients. Four scrims are free, including the first challenge.
Intro to AI Engineering
Level: Intermediate | Duration: 2.5 hrs | Access: Pro
LLM API basics on a Gift Genie app: the messages array and system prompt, streaming with for await, structured outputs, then a super challenge that moves the API key onto a Node server. Project 1 below is its first half. Five scrims free; the PollyGlot solo project is Pro.
Build a Support Agent with the Vercel AI SDK
Duration: 1.9 hrs | Access: Pro
Supabase as a vector store, then generateObject with Zod, tool calling with stopWhen, a routing agent over nine Scrimba help-center articles, and web search, behind an Express route with a chat page. SDK v5, so older tutorials will not match. Six scrims (32 minutes) free, the other 17 Pro.
AI engineering practice projects
Project 1: Build a chatbot with memory
A chatbot on the OpenAI API that remembers the conversation. Practice: API calls, message arrays, system prompts, streaming.
Project 2: Build a RAG system
Answer questions about your own documents. Practice: chunking, embeddings, vector similarity search, prompt augmentation.
Project 3: Build an AI agent with tools
Create an agent that can search the web, read files, and execute code. Practice: tool definitions, agent loops, and error handling.
Project 4: Build an MCP server
Create a Model Context Protocol server that exposes your data to AI assistants. Practice: protocol implementation, tool definitions, and resource exposure.
How to practice AI engineering effectively
- Start with APIs. Call OpenAI, Anthropic, or Mistral before building complex systems.
- Build incrementally: a basic chatbot, then memory, then tools, then RAG.
- Test with real data, your own documents, not toy datasets.
- Deploy something, so you can show it to employers or clients.
Who should start here
Start here if you want concrete course picks for hands-on AI practice, not a syllabus. Most of what matters sits behind Pro.
Related practice guides
- Practice API Calls, essential for working with LLM APIs
- Practice JavaScript Arrays, data manipulation for AI pipelines
- Practice TypeScript, type-safe AI application development
Related pages
- AI & ML Courses | AI Engineer Path
- All Courses | Scrimba Pricing
- Scrimba vs Educative, AI course comparison
No. AI engineering focuses on using language models through APIs, not building them from scratch. If you know JavaScript and can make API calls, you can start learning AI engineering today.
Intro to Mistral AI, Build Serverless AI Agents with Langbase, and the ChatGPT for Programming course are free. The AI Engineer Path and specialized courses like Learn RAG and Learn AI Agents require Scrimba Pro.
Yes. Demand for developers who can integrate AI into products has risen sharply, and the skills here, API integration, prompt engineering, RAG, agents, apply directly to real job roles.
Ship AI features with guided scrims
Try free AI intros on Scrimba; unlock the full AI Engineer Path with Pro.