Summary

This project addresses the gap between superficial AI tool usage and deep technical understanding by offering a comprehensive, hands-on curriculum for AI engineering. It targets engineers, researchers, and power users who aim to build AI systems from first principles, providing a structured, 20-stage path to mastering AI development. The core benefit is enabling learners to construct and understand AI components deeply, moving beyond API-level interaction.

How It Works

The project provides a 503-lesson curriculum focused on building AI from scratch, emphasizing foundational mathematics and algorithm implementation before introducing frameworks. It spans 20 stages, covering topics from linear algebra to autonomous agent clusters, using Python, TypeScript, Rust, and Julia. Each lesson follows a cycle of problem understanding, mathematical derivation, coding, testing, and artifact generation (prompts, skills, agents, or servers), eschewing superficial "quick-fix" approaches.

Quick Start & Requirements

• Primary Access: Lessons can be read directly on the dedicated Chinese website aieng-zh.cn without any setup.
• Code Execution: Users can clone the repository and run provided Python examples (e.g., python phases/01-math-foundations/01-linear-algebra-intuition/code/vectors.py).
• Prerequisites: Basic programming skills (Python proficiency recommended) and a strong desire to understand AI internals are required.
• Interactive Learning: Built-in agent commands like /find-your-level offer personalized learning paths and assessments.
• Relevant Links: aieng-zh.cn, CONTRIBUTING.md, ROADMAP.md.

Highlighted Details

• Localized Content: Features full Simplified Chinese translation for all course materials, including diagrams and quizzes, while retaining key technical terms in English.
• Interactive Website: aieng-zh.cn offers a searchable course catalog, progress tracking, interactive charts, and a command palette.
• Visual Explanations: Accompanying 3Blue1Brown-style animated videos with Chinese voiceovers (Phase 1 complete, others in progress) visually explain mathematical derivations.
• Reusable Artifacts: Each lesson delivers a tangible output, such as prompt templates, skills for AI agents, or Model Context Protocol (MCP) servers.
• Comprehensive Curriculum: Covers 20 stages from foundational math and ML to advanced topics like LLMs, multimodal AI, agent engineering, and production infrastructure.
• Tooling & Automation: Includes scripts for installing agent skills, scaffolding agent workbenches, and automated validation of course content consistency.

Maintenance & Community

The project is maintained by Rohit Ghumare and the community. Key resources for contribution and community engagement include CONTRIBUTING.md, FORKING.md, ROADMAP.md, and CODE_OF_CONDUCT.md.

Licensing & Compatibility

Released under the permissive MIT License, this project allows for extensive use, modification, and distribution, including commercial applications, with minimal restrictions.

Limitations & Caveats

Animated video content is still under development for later course phases. The "from scratch" pedagogical approach necessitates a significant time commitment, estimated at over 320 hours for the full curriculum. Technical terms are often retained in English within the Chinese text.