Recommended AI Courses & Learning Paths
A curated list of reliable, trusted online courses for AI engineers, ML practitioners, and product teams, chosen for concepts that age well. No fluff, no outdated MOOCs. Providers rename, move, and retire courses, so if a link redirects, follow it to the provider's current version.
Note on code staleness: these courses are curated for their concepts. The framework APIs they use (LlamaIndex, LangChain) churn fast, so notebook code in any recorded course drifts out of date and can fail on a fresh install. Before copy-pasting, run the 30-second currency check in Navigating Framework Churn, and learn the primitives, not the import paths.
Foundation: LLMs & Transformers
| Course | Provider | Cost | Why It's Trusted |
|---|---|---|---|
| Neural Networks: Zero to Hero | Andrej Karpathy (YouTube) | Free | The definitive from-scratch series by an OpenAI/Tesla legend. Builds GPT from scratch. |
| CS324: Large Language Models | Stanford | Free | Stanford-quality lecture notes covering LLM fundamentals, scaling laws, alignment. |
| Generative AI with LLMs | DeepLearning.AI + AWS (Coursera) | ~$50 | Hands-on intro to LLMs, covering training, fine-tuning, RLHF. By Andrew Ng's team. |
| Practical Deep Learning for Coders | fast.ai | Free | Bottom-up, code-first approach. Best for engineers who learn by doing. |
RAG Pipelines
| Course | Provider | Cost | Why It's Trusted |
|---|---|---|---|
| Building and Evaluating Advanced RAG | DeepLearning.AI + LlamaIndex | Free | Covers sentence-window retrieval, auto-merging, RAG evaluation with TruLens. |
| Vector Databases: from Embeddings to Applications | DeepLearning.AI + Weaviate | Free | Practical walkthrough of embeddings, vector stores, and hybrid search. |
| Building RAG Agents with LLMs | NVIDIA Deep Learning Institute | Free | Enterprise-grade RAG with NVIDIA NIM. Covers chunking, reranking, evaluation. |
| LlamaIndex: Understanding LlamaIndex | LlamaIndex | Free | Official LlamaIndex learning path: short tutorials on agents, workflows, and RAG, plus observability and evaluation. Best for deep RAG pipeline mastery. |
| RAG Fundamentals (Haystack) | deepset / Haystack | Free | Hands-on tutorials for pipeline-based RAG using the Haystack framework. |
Agentic AI & Multi-Agent Systems
| Course | Provider | Cost | Why It's Trusted |
|---|---|---|---|
| AI Agents in LangGraph | DeepLearning.AI + LangChain | Free | LangGraph by the creators. Covers ReAct, persistence, human-in-the-loop, multi-agent. |
| Multi AI Agent Systems with crewAI | DeepLearning.AI + crewAI | Free | Official CrewAI course. Covers Crews, Flows, and real-world business automations. |
| Building Agentic RAG with LlamaIndex | DeepLearning.AI + LlamaIndex | Free | Routing, tool-calling agents, and multi-document agentic retrieval. |
| Functions, Tools and Agents with LangChain | DeepLearning.AI + LangChain | Free | Tool-calling, OpenAI function calling, building from scratch. |
| AI Agentic Design Patterns with AutoGen | DeepLearning.AI + Microsoft Research | Free | Multi-agent patterns: debate, tool use, and code execution agents. AutoGen itself is now in maintenance mode, succeeded by Microsoft Agent Framework (GA April 2026), so take the patterns, not the API. |
| CS294/194-196: LLM Agents (Berkeley) | UC Berkeley | Free | Graduate-level course on LLM agents. Covers memory, planning, safety, evaluation. |
| Model Context Protocol Associate (MCPA) | Linux Foundation / Agentic AI Foundation | $250 (exam) | A certification, not a course: a proctored, 90-minute multiple-choice exam aligned with MCP spec 2026-07-28, launched September 14, 2026. Security and governance is 24% of it. A vendor-neutral signal, but shipped MCP work still counts for more. |
Context & Memory Management
| Course | Provider | Cost | Why It's Trusted |
|---|---|---|---|
| Building Systems with the ChatGPT API | DeepLearning.AI + OpenAI | Free | Covers multi-turn conversation state, context management, moderation chains. |
| Prompt Engineering with Llama 2 | DeepLearning.AI + Meta | Free | Shows context window tradeoffs and system prompt management with Llama 2. |
| Reasoning with o1 | DeepLearning.AI + OpenAI | Free | Prompting and planning with reasoning models. The model is older than anything you will deploy, but the ideas map directly onto today's reasoning-effort controls (reasoning.effort on OpenAI, adaptive thinking with an effort setting on Claude). |
| Mem0 Documentation | Mem0 (official) | Free | Official reference for multi-tier memory in production agents. Mem0 2.x changed the write model (memories are add-only since 2.0) and current releases drop graph drivers, so older tutorials break. |
AI Evaluations & Observability
| Course | Provider | Cost | Why It's Trusted |
|---|---|---|---|
| AI Evals For Engineers & PMs | Hamel Husain & Shreya Shankar (Maven) | Paid (cohort) | Industry gold standard for AI evals. Used in production at dozens of companies. Our repo's evals guides are based on this course. |
| Evaluating and Debugging Generative AI | DeepLearning.AI + W&B | Free | Covers tracing, evaluation with W&B Weave, and experiment tracking. |
| Quality and Safety for LLM Applications | DeepLearning.AI + WhyLabs | Free | Covers hallucination detection, toxicity, bias evaluation, and drift monitoring. |
| LangSmith Evaluation Concepts | LangChain | Free | Official LangSmith docs are the best hands-on eval reference if you use the LangChain ecosystem. Covers offline and online evaluation. |
| Arize Phoenix Docs | Arize Phoenix | Free | Hands-on tutorials for tracing and evals with Phoenix, which is source-available under the Elastic License 2.0. Dynatrace completed its acquisition of Arize on October 1, 2026 and has said Phoenix remains available. |
📖 Also see this repo's companion guides:
Prompt Engineering & Context Engineering
| Course | Provider | Cost | Why It's Trusted |
|---|---|---|---|
| ChatGPT Prompt Engineering for Developers | DeepLearning.AI + OpenAI | Free | The foundational prompt engineering course. By Isa Fulford & Andrew Ng. |
| Interactive Prompt Engineering Tutorial | Anthropic (GitHub) | Free | Straight from the Claude team: nine chapters with exercises covering prompt structure, XML tags, separating data from instructions, examples, and avoiding hallucinations. It runs on an older Claude model, but the techniques transfer. |
| DSPy: Building Optimizable Pipelines | Stanford NLP (GitHub) | Free | Not a course but the DSPy repo's notebooks are the best way to learn programmatic prompting. |
| Prompt Engineering Guide | DAIR.AI | Free | Comprehensive, community-maintained reference covering all major prompting techniques. |
Fine-tuning & Adaptation
| Course | Provider | Cost | Why It's Trusted |
|---|---|---|---|
| Finetuning Large Language Models | DeepLearning.AI + Lamini | Free | Covers LoRA, full fine-tuning, dataset preparation, evaluation. Concise and practical. |
| Reinforcement Learning from Human Feedback | DeepLearning.AI + AWS | Free | Deep dive into RLHF: reward models, PPO, preference datasets. |
| Hugging Face LLM Course | Hugging Face | Free | Formerly the NLP Course. The best free course for fine-tuning transformers with the HF ecosystem (Trainer, PEFT, etc). |
| How Diffusion Models Work | DeepLearning.AI | Free | For image model fine-tuning (stable diffusion, LoRA for images). |
Inference Optimization & MLOps
| Course | Provider | Cost | Why It's Trusted |
|---|---|---|---|
| Machine Learning in Production | DeepLearning.AI (Coursera) | Paid | Andrew Ng's course on the production ML lifecycle: deployment patterns, monitoring, and data-centric fixes. The older four-course MLOps specialization now redirects here. |
| Efficiently Serving LLMs | DeepLearning.AI + Predibase | Free | Covers vLLM, PagedAttention, quantization, LoRA serving. Exactly what this guide covers. |
| vLLM Documentation & Tutorial | vLLM | Free | The official vLLM docs are the most up-to-date reference for high-throughput serving. |
AI Safety & Guardrails
| Course | Provider | Cost | Why It's Trusted |
|---|---|---|---|
| Red Teaming LLM Applications | DeepLearning.AI + Giskard | Free | Hands-on red teaming, prompt injection, jailbreak detection, bias testing. |
| AI Alignment | BlueDot Impact | Free | Most trusted free course on AI alignment and safety (formerly AI Safety Fundamentals). Used by professionals at Anthropic, DeepMind. |
| NeMo Guardrails | NVIDIA | Free | Hands-on notebooks for building production guardrails with NeMo Guardrails. |
Coding Agents & Developer AI Tools
| Course | Provider | Cost | Why It's Trusted |
|---|---|---|---|
| Claude Code: Official Docs | Anthropic | Free | The definitive starting point for Claude Code. Covers CLAUDE.md (and reading an existing AGENTS.md), skills, hooks, subagents, the Agent SDK, and permissions. |
| Claude Code in Action | Anthropic (Claude Academy) | Free | Short self-paced course (9 lessons, about an hour) on long, hands-off sessions: steering, configuration, automation, and verifying unsupervised runs. |
| Building Code Agents (Hugging Face) | Hugging Face | Free | HuggingFace's official agents course, which includes a unit on building code-execution agents with smolagents. |
| OpenHands Documentation | OpenHands | Free | Official docs for the OpenHands autonomous coding agent: the browser canvas, the CLI, and the Software Agent SDK. |
📖 Also see this repo's guides: Claude Code Guide and OpenCoder Landscape
For PMs & Non-Engineers
These require no Python experience:
| Course | Provider | Cost | Why It's Good |
|---|---|---|---|
| AI for Everyone | DeepLearning.AI (Coursera) | Free | Andrew Ng's course for non-technical roles. Covers what AI can/can't do, project leadership. |
| Prompt Engineering for Everyone | Learn Prompting | Free | Plain-English guide to prompt engineering for non-engineers. |
| AI Evals For Engineers & PMs (Maven) | Hamel Husain & Shreya Shankar | Paid | Despite having code, this course is designed for PMs and QAs, not just engineers. Highly recommended. |
| AI Product Manager: Real Role or Buzzword? | Product School | Free (blog) | Practical guide to what an AI PM actually does, how it differs from a PM who uses AI tools, and the skills to build. |
| Google: Introduction to Generative AI | Google Cloud Skills Boost | Free | No-code introduction to generative AI, LLMs, and responsible AI. |
YouTube Channels & Free Content
| Channel / Resource | Focus | Why Follow |
|---|---|---|
| Andrej Karpathy | Foundations, transformers | Best explanations of how LLMs actually work |
| Yannic Kilcher | Paper reviews | Clear walkthroughs of latest ML research papers |
| Aleksa Gordić - The AI Epiphany | Paper reviews | Deep technical paper breakdowns |
| AI Jason | Agents, LangChain, practical | Great intro videos for agentic frameworks |
| Sam Witteveen | Gemini, RAG, agents | One of the best practical AI YouTubers |
| Matt Wolfe | AI news, product demos | Best for staying current on AI news and tools |
| Hamel Husain (blog) | Evals, production AI, LLMs | Real production insights from the author of the evals maven course |
| Simon Willison (blog) | LLM news, tools, coding | The most trustworthy daily AI news source |
| The Latent Space podcast | Technical AI interviews | Best technical AI podcast, with deep dives with researchers |
| Lex Fridman Podcast | Broad AI/ML interviews | Long-form interviews with leading AI researchers |
Learning Path Suggestions
🛤️ Path: "I'm new to AI and want to build things fast"
Week 1: Prompt Engineering for Developers (DeepLearning.AI): free, 2 hrsWeek 2: Building Systems with ChatGPT API (DeepLearning.AI): free, 2 hrsWeek 3: Building and Evaluating Advanced RAG (DeepLearning.AI): free, 2 hrsWeek 4: AI Agents in LangGraph (DeepLearning.AI): free, 4 hrsMonth 2: Pick a real project, use this guide as reference
🛤️ Path: "I want to understand LLMs deeply"
Week 1-3: Neural Networks: Zero to Hero (Karpathy): free, 12+ hrsWeek 4-6: CS324 Stanford LLMs: free, 30+ hrsMonth 2: Generative AI with LLMs (Coursera DeepLearning.AI)Month 3: CS294 LLM Agents (Berkeley)
🛤️ Path: "I want to build production-ready AI evaluation"
Week 1: Evaluating and Debugging Generative AI (DeepLearning.AI + W&B): freeWeek 2: This repo's evals guides (Phoenix/Langfuse): free ← start hereWeek 3-4: Quality and Safety for LLM Applications (DeepLearning.AI)Month 2: AI Evals For Engineers & PMs (Maven, Hamel + Shreya): paid, worth it
🛤️ Path: "I'm a PM learning to contribute to AI product quality"
Week 1: AI for Everyone (Coursera): freeWeek 2: Prompt Engineering for Everyone (learnprompting.org): freeWeek 3: AI Evals guide in this repo: free (especially Chapters 1-3 on error analysis)Month 2: AI Evals For Engineers & PMs (Maven): paid, has PM track
🛤️ Path: "I want to deploy coding agents in my team"
Day 1: Claude Code docs (code.claude.com): freeWeek 1: This repo's Claude Code Guide + OpenCoder Landscape GuideWeek 2: Claude Code in Action (Claude Academy) + Building Code Agents (Hugging Face): freeMonth 1: Run Claude Code on a real project in CI
How to Stay Current
AI moves fast. Beyond courses, these habits keep you current:
- Follow Simon Willison's blog: daily, trustworthy AI news summaries
- Read Anthropic + OpenAI release notes: primary sources beat second-hand summaries
- Watch the Latent Space podcast: best technical depth
- Contribute to open source (OpenHands, LlamaIndex, DSPy): real learning happens in PRs
- Star this repo: we update it as the landscape changes ⭐
Maintained by Om Bharatiya. PRs welcome for new course additions!
From the AI System Design Guide by Om Bharatiya · MIT License ·view source