itshen/learn-ai

What it solves

Most AI learning content is scattered, paywalled, or too code-heavy for product managers. This project gathers a complete, structured curriculum that walks a non-engineer from the underlying principles of large language models to practical AI engineering and productization — all in plain, slide-based lessons and free to use.

How it works

The course is organized into seven major chapters plus a bonus dessert section, totaling 54 topics and more than 180 pages of interactive slides. Each slide is a single HTML file stored under the slides/ directory — you simply open any file in a browser (start with slides/home.html or slides/learn.html). No server or build tooling is required.

Topics progress from token generation and model training, through prompt engineering, Agent design, vector databases, RAG pipelines, and cost optimization, then move into real-world production cases, engineering design patterns, and even a deep dive into the Rust source code of a production coding agent. The final chapters cover vibe coding methodology and a bonus startup course distilled from Lei Jun's public lectures.

Who it's for

This is aimed at AI product managers, independent developers, and anyone who wants to understand AI systems from a product and engineering perspective without getting lost in dense academic papers or low-level code. It is also useful for founders or team leads who need practical knowledge about AI productization, agent architectures, and cost management.

Highlights

• Completely free and open source (AGPL-3.0) — the author explicitly chose to remove paywalls to spread AI knowledge
• 7 main chapters + 1 bonus section, 54 topics, over 180 interactive slides
• Self-contained HTML slides — just open in a browser, no installation or server needed
• Covers both theory (tokenization, hallucination causes and mitigations) and practice (productizing demo apps, multi-agent collaboration, MCP ecosystem, cost reduction with KV cache)
• Includes real-world code walkthroughs, e.g., dissecting a production coding agent built in Rust
• Bonus material: Lei Jun's startup methodology (fundraising, valuation, equity, cash-flow management)