Zero-to-One AI Learning Roadmap: From Python to Your First Project

Who this roadmap is for

Readers with no programming or math background who want to learn AI systematically. The roadmap relies mostly on free resources and takes roughly 4–6 months at 2 hours per day to reach your first project. If you already code, start from stage three.

Stage 1: Python foundations (3–4 weeks)

Nearly all AI tooling and tutorials use Python, so clear the language hurdle first:

• Runoob Python 3 — a fast Chinese-language syntax reference.
• The Official Python Tutorial — authoritative and free; read it as your second pass.
• MofanPy Python Course — video lectures in Chinese.

Focus on variables and control flow, functions, lists and dicts, file I/O, and pip. You don't need advanced OOP on day one.

Stage 2: Math refresher (2–3 weeks)

You don't need to become a mathematician, but linear algebra, probability and statistics will catch up with you:

• StatQuest — the most intuitive explanation of statistics concepts anywhere.
• Mathematics for Machine Learning (Coursera, Imperial College London) — systematic linear algebra and calculus.

Stage 3: Machine learning (5–6 weeks)

• Hung-yi Lee's Machine Learning Course (NTU) — the de-facto standard Chinese ML intro course.
• scikit-learn User Guide — run regression, classification and clustering with your own hands.
• MIT 6.036 Intro to Machine Learning — a great English-language alternative.

Stage 4: Deep learning and specialization (5–6 weeks)

• PyTorch Official Tutorials — the mainstream deep learning framework, from tensors to training loops.
• MIT 6.S191 Intro to Deep Learning — an intense one-week course for the big picture.
• Then specialize: LLMs and RAG in our Large Language Models and RAG categories, coding tools in AI Coding.

Stage 5: Your first real project (3–4 weeks)

Projects are the only real test of learning. Start by imitating:

• Project Based Learning — a curated list of tutorial projects by language and stack.
• Kaggle Competitions — begin with Titanic and study public notebooks.

Three pieces of advice

1. Don't hoard tutorials: finishing one line of study beats bookmarking ten courses.
2. Code early: understanding and writing are separated by dozens of small bugs.
3. Accept forgetting: looking things up again is normal — that's what the roadmap is for.