
Learning Roadmaps
AI Engineer
Go from Python basics to building real AI systems: understand how LLMs work, master prompting, then build RAG pipelines and AI agents step by step.
Roadmap badge
Complete all courses in this roadmap to earn its badge.
Courses in this roadmap
- Step 1Beginner~4h
Python for Beginners
Learn Python from scratch by building 8 mini-projects you run in your browser, from a tip calculator to a flashcard quiz app. No install, no signup to start, and a free certificate. Powered by Pyodide.
- Step 2Beginner40 min
How LLMs Actually Work: Tokens, Parameters, and MoE Explained
Go one level under the hood of ChatGPT, Claude, and Gemini. Understand tokens, next-token prediction, context windows, model parameters, Mixture of Experts, and why AI models cost money to run. A short, no-code course for people who use AI every day.
- Step 3Beginner~5h
Prompt Engineering Course: Master AI Prompts with Hands-On Practice
Master the art of crafting effective AI prompts through hands-on exercises. Learn prompt structure, few-shot learning, chain-of-thought reasoning, and advanced techniques with instant feedback. Build real-world prompts for code generation, content writing, and data analysis.
- Step 4Intermediate
Vector Databases: The Foundation of AI Apps
Master vector databases for AI applications. Learn embeddings, similarity search, and hands-on setup of Pinecone, pgvector, and Chroma. Understand indexing strategies, hybrid search, performance optimization, and how to choose the right database for your use case.
- Step 5Beginner75 min
Local RAG for Beginners: Build a Private Knowledge Base
Build a private, offline knowledge base over your own notes and PDFs using a local AI model. A hands-on beginner micro course: run a model with Ollama, turn documents into embeddings, store them in a local Chroma database, and ask questions answered from your own files. No cloud APIs, no fees, your data stays on your machine. Want to deploy it as a web app instead? See the Full-Stack RAG course (/courses/fullstack-rag-nextjs-supabase-gemini).
- Step 6Beginner35 min
Build Your First AI Agent in 30 Minutes
Build a working AI agent from scratch in Python. Learn the think-act-observe loop, call an LLM, give your agent a tool it can use, and handle a failed tool call. Framework-agnostic and beginner-friendly.
- Step 7Intermediate
Agentic AI with Python — LangChain & LangGraph
Build autonomous AI agents with Python using LangChain and LangGraph. Learn tool calling, stateful workflows, RAG-powered agents, multi-agent systems, and production deployment. From the ReACT pattern to a full customer support agent capstone.
Practice projects
Prove what you learned by building something real. Pick a project, build it on your own machine, push it to a public GitHub repository, and submit the link for review.
AI Research Assistant
Build a tool that accepts a research question, searches external sources, and uses an LLM to produce a structured answer with citations.
Requirements
- Accepts a user question from the command line or a simple UI
- Searches at least one external source (web search API, Wikipedia, or arXiv)
- Uses an LLM API to synthesize the findings
- Cites the sources it used in the final answer
- Produces structured output (sections or JSON, not one blob of text)
- Handles errors: bad input, failed requests, empty search results
Suggested tools: Python or JavaScript, An LLM API, A search API
Chat With Your Documents
Build a small RAG pipeline: load your own documents, embed them into a vector store, and answer questions grounded in what the documents actually say.
Requirements
- Loads a folder of text, Markdown, or PDF documents
- Chunks the documents and stores embeddings in a vector store
- Retrieves relevant chunks for each question
- Answers using only the retrieved content and says which document it came from
- Says "I don't know" when the documents don't cover the question
- README explains how to run it with your own documents
Suggested tools: Python, An embeddings model, A vector store such as Chroma or FAISS
AI Agent That Completes a Task
Build an agent that plans and executes a multi-step task using tools, for example researching a topic and saving a report, or triaging files in a folder.
Requirements
- The agent has at least two tools it can call (search, read file, write file, call an API)
- Runs a reasoning loop: decide, act, observe, repeat until done
- Logs each step so a reviewer can follow what it did
- Has a stop condition and a maximum step limit
- README explains the task, the tools, and one example run
Suggested tools: Python, LangChain or a hand-written agent loop, An LLM API

