DeepLearning.AI Complete Course Guide: 120 Courses, Taught by the Authors Themselves — 6 Goal-Based Learning Paths

DeepLearning.AI now has 120 courses — more than Anthropic, OpenAI, Google, NVIDIA, Microsoft, AWS, Meta, and IBM combined.

But the number isn't the point. What this platform has that nobody else can match is:

The courses are taught directly by the people who built the models and tools.

Anthropic: 4 courses
OpenAI: 4 courses
Meta: 4 courses
Hugging Face: 5 courses
LangChain: 5 courses

Each course runs 1–2 hours, is completely free, and requires no credit card. Getting taught how to use a tool by the person who built it is something no secondhand tutorial can replicate.

Start here: deeplearning.ai/courses

Below, I've broken everything down into 6 goal-based learning paths — choose by goal, not by popularity.

Goal 1 · Prompting and AI Workflows

AI Prompting for Everyone (Andrew Ng)
ChatGPT Prompt Engineering for Developers (OpenAI)
Generative AI for Everyone (Andrew Ng)

These three courses are the foundation for going from zero to confidently integrating AI into your everyday workflow.

Goal 2 · Building Agent Systems

Agentic AI (Andrew Ng — structured course series)
MCP: Build Rich-Context AI Apps with Anthropic
Agent Skills with Anthropic
AI Agents in LangGraph
Building AI Browser Agents
Building toward Computer Use with Anthropic

Agents are the hottest direction heading into 2026 — this is the longest path on the list and the one most worth investing in. The two Anthropic official courses, MCP and Agent Skills, are especially critical.

Goal 3 · Building RAG Systems

Retrieval Augmented Generation (RAG) — structured course series
Building Agentic RAG with LlamaIndex
Knowledge Graphs for RAG (Neo4j)
Advanced Retrieval for AI with Chroma

RAG remains the most widely used architecture in enterprise AI applications. This path covers everything from foundational RAG to knowledge graph-powered RAG.

Goal 4 · Fine-Tuning Models

Fine-tuning & RL for LLMs: Intro to Post-training (AMD)
Post-training of LLMs (University of Washington / NexusFlow)
Reinforcement Fine-Tuning LLMs With GRPO

If you want to move beyond "calling the API" to "modifying the model itself," this is the path to take. The GRPO course covers the most cutting-edge RL fine-tuning methods in use today.

Goal 5 · Building a Deep Learning Foundation

Machine Learning Specialization (with Stanford)
Deep Learning Specialization
Mathematics for Machine Learning and Data Science

If your goal isn't just "knowing how to use it" but "understanding how it actually works," these three Specializations are the gold standard.

Goal 6 · AI-Assisted Programming

Build with Andrew (build an app in 30 minutes)
Vibe Coding 101 with Replit
Generative AI for Software Development
Spec-Driven Development with Coding Agents

From zero-to-app vibe coding all the way to spec-driven development with coding agents — this path covers the entirely new programming paradigm of the AI era.

My Learning Recommendations

Choose by goal, not by popularity: The 6 paths map to 6 different objectives — get clear on what you want before you start
Prioritize courses taught by the creators: Anthropic / OpenAI / Meta / Hugging Face / LangChain official courses are first-hand knowledge you can't get elsewhere
Focus on one path at a time: With 120 courses, the biggest risk is overcommitting — stick to one goal for three months
Subscribe to the short course list: Create an account and subscribe — new courses typically show up in the list 24–48 hours before the official announcement
Pair with courses from the major platforms: DeepLearning.AI leans toward "applied + cutting-edge," while the big-company courses lean toward "certification + ecosystem" — they complement each other well

FAQ

Are DeepLearning.AI courses free?

All 120 short courses are completely free — no credit card required. Each runs 1–2 hours, and you can start learning as soon as you create an account.

What advantage does DeepLearning.AI have over official courses from major tech companies?

The tools are taught by the people who built them: Anthropic (4 courses), OpenAI (4 courses), Meta (4 courses), Hugging Face (5 courses), LangChain (5 courses). Having the builders teach you how to use their own tools is first-hand knowledge that no secondhand tutorial can replicate.

With 120 courses, where should I start?

Choose by goal, not by popularity. For everyday productivity, start with AI Prompting for Everyone (Andrew Ng). For building agents, go with Agentic AI plus the two Anthropic official courses on MCP. For RAG, start with the RAG structured course series. Focus on one path at a time.

The Bottom Line

DeepLearning.AI's real advantage isn't having 120 courses — it's that the people who built the tools are the ones teaching you how to use them. Choose a path based on your goal: prompting, agents, RAG, fine-tuning, deep learning fundamentals, or AI-assisted programming. Every path is first-hand knowledge. It's all free, no credit card required — and it's the highest-information-density resource in the entire AI free learning landscape.

→ IBM SkillsBuild Free AI Courses Guide → Meta PyTorch Official Tutorials Guide → AWS Skill Builder Free AI Courses Guide → Microsoft AI Certification Complete Guide → NVIDIA DLI Free Courses Complete Guide → Google Skills Gamified Learning Paths → Claude Certified Architect Complete Roadmap → OpenAI Academy vs. Anthropic: Full Comparison

Next up: Hugging Face — the platform where anyone serious about open-source LLMs starts their learning journey.

📚 The Complete AI Free Learning Map Across 10 Platforms → One table + role-based learning paths — the full series index.