A.6 Frequently Asked Questions

Use this page when the question is emotional or vague. Convert it into a next action.

Fast answers

Question	Short answer	Next action
I am weak at math. Can I learn AI?	Yes. Learn math through code and projects.	Run Python, NumPy, and small ML examples first.
Do I need a GPU?	Not at the beginning.	Use CPU/API first; rent cloud GPU only when training needs it.
Do I need every chapter?	Keep the main path, then choose direction modules.	Finish foundations, then pick RAG/Agent/CV/NLP/multimodal focus.
How many hours per week?	Consistency beats bursts.	4-10 hours/week is enough if sustained.
When should I do projects?	As early as possible, but keep them small.	Build one input -> one process -> one output.
I cannot read papers. Is that fatal?	No. Papers are later supplements.	Read tutorial, code, then paper.
When can I job hunt?	When you can explain 2-3 projects clearly.	Prepare README, metrics, failure cases, and interview story.

Minimum learning path for application builders

Tools -> Python -> Data -> Deep Learning Basics -> LLM Principles -> RAG -> Agent

If you want stronger model foundations:

Tools -> Python -> Data -> Math -> ML -> Deep Learning -> LLM -> RAG / Agent

Weekly rhythm

Weekly time	Good rhythm
4-6 hours	2 study sessions + 1 coding session
7-10 hours	3 study sessions + 2 coding sessions
12-18 hours	Add one project/review block, avoid burnout

Confidence reset

When you feel slow, ask:

1. Can I understand more code than last month?
2. Can I modify examples instead of only copying?
3. Can I explain one concept more clearly?

If yes, you are progressing.

Three abilities to strengthen first

1. Python and debugging.
2. Data processing and visualization.
3. Minimal closed-loop project thinking.

Many beginners think they need a more advanced model. Often they first need clearer inputs, outputs, checks, and explanations.