AI ENGINEERING · COURSE NOTES

从问题定义到可靠的 AI 系统

围绕检索增强生成、智能体与多智能体协作，学习数据处理、系统评估、部署与持续优化。

开始阅读

环境准备

环境准备

1 课
  1. 01搭建开发环境Setting up your development environment

Sprint 0

Sprint 0 – 问题定义、基础设施搭建与 RAG 基础

10 课
  1. 01理解 AI 产品生命周期Understanding the AI product lifecycle
  2. 02工具概览：LangGraph、向量数据库与 LLM APITooling overview (LangGraph, vector DBs, LLM APIs, etc.)
  3. 03什么是 RAG？What is RAG
  4. 04嵌入模型与向量数据库集成Embedding models & vector DB integration
  5. 05建立基础可观测性能力Implementing basic observability foundations
  6. 06评估端到端检索与生成Evaluating basic end-to-end retrieval and generation
  7. 07实践环节Hands-On Section
  8. 08Amazon 电子产品类别数据集概览Amazon Electronics Category Dataset Overview
  9. 09AI 项目画布（可选）[OPTIONAL] AI Project Canvas
  10. 10成功指标与评估框架（可选）[OPTIONAL] Success metrics & evaluation frameworks

Sprint 1

Sprint 1 – 检索质量与上下文工程

6 课
  1. 01RAG 数据摄取流水线RAG Data Ingestion Pipeline
  2. 02Pydantic 与结构化输出Pydantic and structured outputs
  3. 03分块策略与 Contextual EmbeddingsChunking strategies and Contextual embeddings
  4. 04上下文工程与提示词管理Context Engineering and prompt management
  5. 05重排序与混合检索Re-Ranking and Hybrid Retrieval for better relevance
  6. 06[可选] 自动化提示词调优(Optional) Automated prompt tuning

Sprint 2

Sprint 2 – 智能体与智能体系统

5 课
  1. 01智能体架构与决策循环Agent architecture and decision loops
  2. 02智能体中的工具调用Tool use in agents
  3. 03构建智能体系统的架构模式Patterns for building Agentic Systems
  4. 04智能体系统中的记忆机制Memory in agent systems
  5. 05反思机制与智能体评估框架Reflection & Agent evaluation Frameworks

Sprint 3

Sprint 3 – 从基础到智能体 RAG

4 课
  1. 01智能体与 RAG 系统的深度集成Agent integrations with RAG systems
  2. 02智能体系统中的人类反馈与容错机制Human Feedback and Fault Tolerance in Agentic Systems
  3. 03人机协同（HITL）Human in the loop (HITL)
  4. 04模型上下文协议（MCP）Model Context Protocol (MCP)

Sprint 4

Sprint 4 – 多智能体系统

4 课
  1. 01什么是多智能体系统以及何时使用Multi-agent systems and when to use it
  2. 02智能体之间的规划、委派与任务路由Planning, delegation, and task routing among agents
  3. 03状态同步与记忆共享Synchronization and memory sharing
  4. 04智能体间通信协议（A2A）Agent-to-agent communication protocols (A2A)

Sprint 5

Sprint 5 – 部署、优化与可靠性

4 课
  1. 01AI 系统的部署架构模式Deployment architecture patterns for AI Systems
  2. 02管理 AI 应用的延迟与成本Managing latency and cost for AI applications
  3. 03构建安全的 AI 系统Securing AI systems
  4. 04AI 应用的持续集成与持续交付（CI/CD）CI CD for AI applications

搜索课程

正在加载课程目录…