A.1 Recommended Learning Resources
Use outside resources only when they solve a current bottleneck. Do not collect links as a substitute for building projects.
Choose by bottleneck
Section titled “Choose by bottleneck”| If you are stuck on | Add this first | Return to |
|---|---|---|
| Concept intuition | 3Blue1Brown, visual explainers, course diagrams | The same chapter exercise |
| Python basics | Python official tutorial, Real Python | Chapter 2 code practice |
| Data work | Pandas, NumPy, Matplotlib, SQL docs | Chapter 3 mini projects |
| Math and ML theory | Andrew Ng ML, Zhou Zhihua, scikit-learn docs | Chapters 4-5 |
| Deep learning code | PyTorch docs, Dive into Deep Learning, Fast.ai | Chapter 6 notebooks |
| LLM usage | Hugging Face docs, model provider docs | Chapters 7-8 |
| RAG / Agent engineering | LangChain, LangGraph, LlamaIndex, MCP docs | Chapters 8-9 projects |
| CV / NLP / multimodal | CS231n, CS224n, OpenCV, diffusion and multimodal docs | Chapters 10-12 |
| Open-source LLM deployment | Datawhale Self-LLM, Hugging Face Transformers, vLLM, SGLang, llama.cpp, Ollama | Chapter 13 |
One-resource rule
Section titled “One-resource rule”For each bottleneck, choose one resource and finish one small output:
| Resource session | Minimum output |
|---|---|
| Watched a concept video | Rewrite the idea in 3 sentences |
| Read documentation | Run one official example |
| Read a blog/tutorial | Apply one idea to your course project |
| Read a paper or survey | Write “old problem -> new method -> project impact” |
Evidence to Keep
Section titled “Evidence to Keep”Keep this page’s proof of learning as a small evidence card:
- Learning Gap
- concept, code skill, project skill, paper, or deployment skill to strengthen
- Resource Choice
- one primary resource and why it fits the current bottleneck
- Time Box
- how long to use it before returning to the project
- Risk Check
- collecting resources instead of producing evidence
- Expected Output
- a short resource plan with one artifact to produce after reading
Common mistakes
Section titled “Common mistakes”- Saving many links but not running code.
- Switching resources whenever one paragraph feels hard.
- Reading theory without returning to a project.
- Trying to learn a whole framework before building the smallest demo.
Quick decision
Section titled “Quick decision”Ask three questions before opening another resource:
- What exact problem am I stuck on?
- Will this resource give a different explanation, an official API answer, or a runnable example?
- Which course task will I return to after reading it?
If you cannot answer those three, stay with the main course path.
Pass Check
Section titled “Pass Check”You pass this appendix page when you choose one resource, define one time box, and name the artifact you will produce before returning to the course project.