Read the curriculum

What you will be able to do

Put the courses for your route in an order that works: no course before the knowledge that it needs. A wrong order is one of the most common reasons that learners stop.

Before you startWhat to know and to have readyAbout 25 minutes

What you need to know

• What a primary role and an adjacent option are, from Choose a realistic route: Primary role and adjacent option.

What you need

• Your primary role and adjacent option, under ## My route in notes.md. You wrote them in Choose a realistic route: Guided practice: your choice.
• This lesson adds the heading ## Course order.

What the catalogue tells you

Each course in the catalogue has a page with:

• Hours: the estimated time, including practice and the final assignment. For example, Python for Practical AI Engineering is 16–22 hours. A beginner may need more.
• Format: what kind of course it is.
• Required courses: the courses that you must complete, or know equally well, before you start.
• Status: some courses are available now. The others are planned, and appear when they are ready.

The order of the table is not a study order. Git comes early in most routes, and Document and multimodal AI comes late.

Format	What it means
Orientation	A first course for everyone, like this one.
Foundation	Basic skills that many courses need: Python, mathematics, SQL.
Core	The main AI topics: machine learning, language models, retrieval, evaluation.
Short practical	One focused skill, such as Git, Docker or releases.
Practical core	A longer hands-on course, such as building and deploying an API.
Advanced	Courses that need many others first, such as fine-tuning models or operating AI systems.
Capstone	A large final project that uses many courses.

All courses and their required courses

This table is the whole planned curriculum, and you use it in the practice below. The catalogue shows the same information: under each planned course, “Take first” names its required courses.

Show all 28 courses

Course	Hours	Required courses
AI Career Foundations: Roles, Skills and Roadmaps	4–6	None
Python for Practical AI Engineering	16–22	None
Essential Mathematics and Statistics for AI	12–16	None
SQL and Data Preparation for AI	12–16	Python
Machine Learning: From Problem to Reliable Model	18–24	Python, Mathematics and statistics, SQL
Deep Learning and Transformers Explained	18–24	Python, Mathematics and statistics, Machine learning
LLMs: How They Work and How to Choose One	8–12	Mathematics and statistics
Building Reliable Applications with LLM APIs	12–16	Python, LLMs
RAG: Building AI That Uses Your Data	16–22	SQL, LLM applications
AI Agents and Workflow Orchestration	14–20	LLM applications, Evaluating AI systems
Evaluating and Testing AI Systems	12–16	Mathematics and statistics, LLM applications
Fine-Tuning and Adapting AI Models	14–20	Deep learning, LLM applications, Evaluating AI systems
Document and Multimodal AI	12–18	LLM applications, RAG
Git and Team Development Essentials	5–7	None
AI-Assisted Software Development	6–8	Python, Git
Build and Deploy Your First API	11–15	Python
Deploying AI Models for Real Users	8–12	Machine learning, First API, Docker and cloud
Docker and Cloud Essentials for AI Developers	7–10	Python, First API
CI/CD, Testing and Safe Releases	8–12	Git, First API, Docker and cloud
Databases and Storage for AI Applications	8–12	SQL, First API
Scaling APIs and AI Workloads	9–12	First API, Docker and cloud, Databases and storage
Frontend Architecture and Performance for AI Apps	12–18	First API
Authentication, Permissions and Multi-Tenant Applications	9–12	First API, Databases and storage
AI Security, Privacy and Responsible Design	10–14	LLM applications, First API, Authentication
MLOps and LLMOps: Operating AI in Production	12–18	Evaluating AI systems, Deploying AI models, CI/CD, Databases and storage, Scaling
AI System Design and Cost Engineering	12–16	RAG, AI agents, Databases and storage, Scaling, Authentication, MLOps and LLMOps
Portfolio Capstone: Build, Evaluate and Ship an AI Product	25–40	LLM applications, RAG, Evaluating AI systems, Git, First API, Docker and cloud, CI/CD, Databases and storage, Frontend, Authentication, AI security
MCP: Connect AI Applications to Tools and Data	12–16	LLM applications, First API, Docker and cloud, Authentication, AI security

Prerequisites make an order

A prerequisite is knowledge that a course needs before it starts. Each course lists its required courses. If you follow the arrows back, you get every course that you need, and an order that works.

Here is the route for Daniela’s primary role from the last lesson: AI product manager, with evaluation as the main technical skill.

The prerequisites for Evaluating AI systems. Solid arrows are required courses. Dotted arrows are a recommended start.

Text description of the diagram

• AI Career Foundations is a recommended start (dotted arrows to Python and to Mathematics and statistics). It is not required by any course.
• LLMs requires Mathematics and statistics.
• LLM applications requires Python and LLMs.
• Evaluating AI systems requires Mathematics and statistics and LLM applications.

To find the order, start at the target and go back:

1. Evaluating AI systems needs Mathematics and statistics and LLM applications.
2. LLM applications needs Python and LLMs.
3. LLMs needs Mathematics and statistics.
4. Python and Mathematics and statistics need nothing.

So a working order is: Python, Mathematics and statistics, LLMs, LLM applications, Evaluating AI systems. Mathematics and statistics and Python can also be in the other order, or at the same time, because neither needs the other.

When you may skip a course

You may skip a required course only when you can already do what it teaches. A feeling is not enough. Use evidence, as in Evidence over self-ratings:

1. Open the course page, and read its final assignment.
2. Ask: “Could I do this assignment now, without help?”
3. If yes, do the parts that you are least sure about. If they work, skip the course.

For example, a software developer may skip Python, but should first check that they can do the Python final assignment.

The learning path recommendation does this check with you. It uses your goal and experience to suggest a route, and marks the courses that you may be able to skip.

Projects grow from course to course

The courses reuse two examples: a support-ticket system and a document assistant, the same kinds of product that you met at Larkfield. In the Python course, you clean ticket data. In the API course, you build a service that classifies tickets. Later courses add a trained model, storage, evaluation and deployment.

So each course gives you a working project, and the next course makes it bigger. By the end of a route, you have a project that you can show. You plan that in Module 5.

Optional extensions

Some lessons have stretch work: optional tasks for people who want more. Stretch work is never required to complete a course, and no later course depends on it. If your time is limited, skip it without worry.

Guided practice: order a route

You did the route to Evaluating AI systems above. Now do one yourself.

1. In notes.md, add a heading: ## Course order.
2. The target is RAG: Building AI That Uses Your Data. It requires SQL and LLM applications.
3. Use the table of all courses above to find what SQL and LLM applications require. Go back until you reach courses that require nothing.
4. Write a working order.
5. Add up the hours, using the lower and the upper estimates.

Show the answer

• RAG needs SQL and LLM applications.
• SQL needs Python.
• LLM applications needs Python and LLMs. LLMs needs Mathematics and statistics.

A working order: Python, Mathematics and statistics, SQL, LLMs, LLM applications, RAG. Other orders work too, if each course comes after the courses that it requires. For example, Mathematics and statistics can come first.

Hours: Python 16–22, Mathematics and statistics 12–16, SQL 12–16, LLMs 8–12, LLM applications 12–16, RAG 16–22. Total: 76–104 hours.

Independent variation: do the same for the target course of your own primary role. If you are not sure which course that is, use the learning path recommendation, and check its order yourself.

Common mistake

Starting with the exciting course

Symptom: you start with the RAG course because it is about building AI that answers questions. In the first lesson, you do not understand the code, the database queries or the words. After a week, you feel that you are “not good enough for AI”.

Cause: RAG needs Python, SQL, and LLM applications. You were not missing talent. You were missing prerequisites.

Fix: follow the arrows back from the course that you want. Start at the first course that you cannot already do. Keep the exciting course as the goal at the end of the plan.

Check your understanding

LLMs requires only Mathematics and statistics. Can you take LLMs before Python? Why might you do that?

Show the answer

Yes. LLMs needs Mathematics and statistics but not Python, so after Mathematics and statistics you can take LLMs before or after Python. A person on a product route might do that, to get AI literacy about language models early, before the programming that LLM applications needs.

Check your understanding

A learner who writes Python at work wants to skip the Python course. What should they do first?

Show the answer

Read the Python course’s final assignment, and check honestly whether they can do it now without help. If they are unsure about some parts, they should try those parts. If the parts work, they skip the course. If not, they take only the lessons that they need. Skipping should come from evidence, not from a job title.

Recap

• Each course has hours, a format, required courses and a status. The table order is not a study order.
• Follow the required courses back from your target to find a working order.
• Skip a course only when you can do its final assignment.
• Projects grow from course to course. Stretch work is optional.

Next, you plan how to study each week, so that the hours become real skill.

Next: Plan deliberate practice