Free AI Roadmap
How to Learn AI From Scratch
A free, step-by-step roadmap from zero to your first AI job, built from what real AI job ads ask for. No experience, no degree and no money needed.
- 224 Lessons in 18 Seasons
- 6 Live Now
- 100% Free
To learn AI from scratch, start with what AI jobs ask for, not with theory: choose a target role, learn Python and data basics, understand how models learn, then build real projects with language models, RAG and agents, and finally learn to deploy, evaluate and secure them. ai2hired teaches this path free, in 224 short video lessons in order.
Start Here: Learn AI to Get Hired: The Full Map
The map of the whole course and the AI job market, in one short video. Then follow the lessons in order.
Open the First LessonThe AI Learning Roadmap, in 8 Stages
Each stage builds on the one before. Every role follows this order, skipping what its job never asks for.
- 1
Start Here 8 Lessons
See the whole AI job market, choose the role you are aiming for, and learn to use AI assistants with judgment from day one.
- Season 0: Start HereUnderstand the AI job landscape, choose a target role, and use AI assistants and prompting with judgment.
- 2
Technical Foundations 32 Lessons
The command line, Python and Git: the everyday tools of every technical AI job, learned with AI assistants at your side.
- Season 1: Computer and Internet FoundationsGet comfortable with Linux and the command line, and understand how software talks over the internet.
- Season 2: PythonRead, write, test and debug Python with confidence, including code written by AI.
- Season 3: Git and GitHubWork like a professional engineer, with version control, code review and automated checks.
- 3
Data and the Math You Need 21 Lessons
Query and clean data, then the few math ideas behind similarity search and model training, explained visually.
- Season 4: Data FoundationsQuery, clean, model, visualize and govern data, and prepare documents for AI.
- Season 5: Math for AIUnderstand the math behind similarity search and model training well enough to reason about models, not just call them.
- 4
How Models Learn 21 Lessons
Classical machine learning for business cases, then neural networks, deep learning and how transformers and large language models work inside.
- Season 6: Classical Machine LearningBuild, evaluate and track machine learning models for the most common business cases.
- Season 7: Deep Learning and TransformersSee how neural networks and large language models work inside.
- 5
Build With Language Models 55 Lessons
Call and combine models in real code, ground them in company data with RAG, build agents that use tools, and wrap it all into a product.
- Season 8: Harnessing Models (CAM)Call, control and combine commercial and open models in production-grade code.
- Season 9: Embeddings, RAG and SearchBuild grounded knowledge systems and search engines on company data.
- Season 10: Agents and Tool UseDesign, build and supervise single and multi-agent systems, and use coding agents professionally.
- Season 11: Building the Product Around the ModelWrap AI into real services, interfaces and enterprise integrations.
- 6
Ship It and Run It 36 Lessons
Deploy to the cloud, prove the system works with evaluation, keep it reliable and affordable, and make it secure and compliant.
- Season 12: Deployment and CloudShip AI systems to the cloud and to regulated environments, repeatably.
- Season 13: LLMOps: Evaluation, Observability, CostProve an AI system works, keep it reliable and keep its cost under control.
- Season 14: Security, Safety and GovernanceMake AI systems secure, compliant and governable, including in regulated industries such as pharma.
- 7
Go Further 37 Lessons
Fine-tune and train models, or lead AI inside a business: strategy, product and transformation.
- Season 15: Customizing and Building ModelsAdapt, optimize and train models, up to the research frontier.
- Season 16: AI Business, Product and TransformationFind, justify, deliver and scale AI inside an organization, and lead the people who do it.
- 8
Get Hired 14 Lessons
Turn your skills into proof: portfolio, CV, visibility, interviews and a signed offer.
- Season 17: Getting HiredTurn your skills into proof, visibility and a signed offer.
What AI Employers Ask For Most
Share of the 2,179 technical job ads at 12 AI companies that name each skill, November 2026. Learn these first.
- 1Python52%
- 2Machine Learning39.8%
- 3Customer-Facing Work28.2%
- 4Large Language Models (LLMs)27.8%
- 5Distributed Systems and Training27.7%
- 6AWS25.1%
- 7AI Agents24.3%
- 8Microsoft Azure21.2%
Learn for a Role, Not for Everything
AI is too wide to learn all of it. Each AI job needs its own part of the roadmap: from 69 lessons for Digital Transformation Manager or AI Transformation Lead to 210 for AI Engineering Manager or Head of AI.
- AI Consultant, AI Business Analyst or Innovation Manager110 Lessons
- AI Engineer190 Lessons
- AI Engineering Manager or Head of AI210 Lessons
- AI Governance or Compliance84 Lessons
- AI Product Manager or AI Product Owner103 Lessons
- AI Research (The Foundation)103 Lessons
- AI Security or Safety Engineer151 Lessons
- AI Workflow or Automation Engineer157 Lessons
- AI, Solution or Enterprise Architect194 Lessons
- Data Engineer for AI111 Lessons
- Data Governance or MDM Lead109 Lessons
- Data Scientist (AI)124 Lessons
- Digital Transformation Manager or AI Transformation Lead69 Lessons
- ML Engineer157 Lessons
- MLOps, LLMOps or AI Platform Engineer159 Lessons
- Solutions Engineer, Forward Deployed Engineer or Technical AI Consultant209 Lessons
Not sure which one? Take the 2-minute Career Quiz or read what each AI job really does.
Mistakes That Slow Down AI Learners
- Learning without a target. Courses on everything feel productive but lead nowhere; a target role tells you what to skip.
- Watching instead of building. Rewatching feels like learning, but only building shows what you can do. The lesson Escape Tutorial Hell: Learn by Building shows how to break the loop.
- Starting with heavy math. Most AI jobs need intuition first; the math comes when a model needs it.
- Ignoring language models. Large Language Models (LLMs) were named in 27.8% of technical AI job ads in November 2026. They are taught early, not as an extra.
- Collecting certificates instead of proof. Employers hire from projects and a CV that matches the ad. Check yours with the free CV Check.
Questions About Learning AI
How do I start learning AI as a complete beginner?
Start by understanding the AI job market and choosing a target role, then learn to use AI assistants well. After that, learn Python, Git and data basics before machine learning. ai2hired Season 0 covers the first step in 8 short lessons, with no code and no math.
Can I learn AI for free?
Yes. Every ai2hired lesson is a free YouTube video with a lesson page; workbooks, tools, career pages and interview questions are free too. 6 lessons are live today, and the full curriculum of 224 lessons is public.
Can I learn AI without coding?
You can understand AI and use it well without coding, and some AI jobs, such as AI product manager, AI consultant or AI governance, need little code. Technical roles like AI engineer need Python, which the course teaches from zero, alongside AI coding assistants.
Do I need Python to learn AI?
For technical AI jobs, yes. Python was named in 52% of the technical job ads at AI companies that ai2hired counted in November 2026, more than any other skill. Season 2 teaches it from zero.
How much math do I need to learn AI?
Less than most people fear. To build with language models you need the ideas behind vectors, similarity and how a model learns from its errors. Season 5 covers them in 6 visual lessons; deeper math matters mainly for research roles.
How long does it take to learn AI?
It depends on the role and your hours. On ai2hired each career path is a fixed list of lessons: the shortest, Digital Transformation Manager or AI Transformation Lead, has 69; the longest, AI Engineering Manager or Head of AI, has 210. Learning a few lessons a week and building as you go, most people see real progress within months, not years.
Is ai2hired only for pharma?
No. ai2hired teaches AI to anyone, for any industry. Many examples come from pharma and life sciences, the course director’s home industry, and every skill works the same everywhere.
What should I learn first in AI?
First, what the AI jobs are and which one you want; second, how to use AI assistants with judgment; third, Python. That order keeps you from spending months on topics your target job never asks for.
Words you do not know yet are explained in the AI Glossary.
Never Miss a Lesson
Subscribe on YouTube, or get an email the day each new lesson goes live.