ai-engineering-from-scratch：AI工程从零开始教程

GitHub 25K星 · 开源教程 · 系统学习AI工程

项目简介

ai-engineering-from-scratch 是 GitHub 上最受欢迎的 AI 工程系统性学习教程之一（25K星），从零开始覆盖 LLM 原理、Prompt 工程、RAG、Agent 开发、模型微调、部署运维等完整技术栈。适合有一定编程基础但刚接触 AI 的开发者。

课程大纲

快速开始

# 克隆仓库
git clone https://github.com/rohitg00/ai-engineering-from-scratch.git
cd ai-engineering-from-scratch

# 安装依赖
pip install -r requirements.txt

# 配置API Key
export OPENAI_API_KEY=sk-xxx
export ANTHROPIC_API_KEY=sk-ant-xxx

# 运行第一章
cd 01-llm-basics
python lesson_01_hello_llm.py

RAG实战示例（第3章）

# 安装RAG依赖
pip install langchain chromadb sentence-transformers

# 构建知识库
from langchain_community.document_loaders import DirectoryLoader
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import Chroma

# 加载文档
loader = DirectoryLoader("./docs", glob="**/*.md")
docs = loader.load()

# 切分
splitter = RecursiveCharacterTextSplitter(
    chunk_size=500, chunk_overlap=50
)
chunks = splitter.split_documents(docs)

# 向量化并存储
embeddings = HuggingFaceEmbeddings(
    model_name="BAAI/bge-small-zh-v1.5"
)
db = Chroma.from_documents(chunks, embeddings)

# 检索问答
results = db.similarity_search("如何部署模型", k=3)
for r in results:
    print(r.page_content[:200])

Agent实战示例（第4章）

# 使用LangGraph构建多Agent
from langgraph import StateGraph, END

def research_agent(state):
    # 搜索信息
    return {"findings": "..."}

def writer_agent(state):
    # 基于发现写文章
    return {"draft": "..."}

def reviewer_agent(state):
    # 审核文章
    return {"approved": True}

graph = StateGraph()
graph.add_node("research", research_agent)
graph.add_node("write", writer_agent)
graph.add_node("review", reviewer_agent)
graph.add_edge("research", "write")
graph.add_conditional_edges("write", "review")
graph.add_conditional_edges("review", END)

推荐配合 LangGraph多Agent编排教程 和 RAG知识库全流程 深入学习具体技术。