Awesome AI Application Engineer

中文 | English

A practical roadmap for AI application engineers: from LLM basics, Prompt Engineering, RAG, Agent, MCP, evaluation, to production deployment.

A hands-on learning repository for developers who want to build reliable LLM applications, not just demos.

This repository brings together learning paths, core concepts, practical tutorials, checklists, templates, bad cases, production notes, and interview preparation. The goal is to help you grow from "I can call a model API" to "I can design, evaluate, deploy, and improve an AI application."

AI application engineer = someone who connects model capabilities to real product and business workflows, while keeping the system stable, observable, controllable, evaluable, and production-ready.

Quick Links

• Online website: Website
• Online docs index: All Pages
• Bilingual tutorial: Build a Personal Knowledge Base Assistant
• Chinese tutorial: 从 0 到 1 构建个人知识库问答助手
• RAG production checklist: RAG Production Checklist
• Agent safety checklist: Agent Safety Checklist
• Prompt review checklist: Prompt Review Checklist
• RAG bad cases: RAG Bad Cases
• Projects: Projects
• Examples: Examples

What This Repository Helps With

• You are new to LLM app development and do not know where to start.
• You understand prompting, but do not know how to build RAG, Agents, or tool calling.
• You can build demos, but do not know how to evaluate, deploy, monitor, and control cost.
• You want to move toward AI application engineering, but need portfolio projects.
• You can explain concepts, but cannot yet describe an end-to-end production system.

Learning Roadmap

flowchart LR
    A["Start Here<br/>Learning Map"] --> A1["Foundations<br/>API, JSON, Data Flow"]
    A1 --> B["LLM Basics<br/>Token, Context, Hallucination"]
    B --> C["Prompt Engineering<br/>Structured Instructions"]
    C --> D["RAG<br/>Knowledge Base Q&A"]
    D --> E["Agent<br/>Tool Calling and Workflows"]
    E --> F["MCP<br/>Connect Tools and Data Sources"]
    F --> G["Evaluation<br/>Eval Sets and Regression"]
    G --> H["Production<br/>Deploy, Monitor, Control Cost"]
    H --> I["Projects<br/>Portfolio and Interview"]

Loading

Learning Path

Stage	Goal	Recommended Content	Output
0. Preparation	Build the learning map and engineering foundation	Start Here, Foundations	Minimal chat demo and data-flow diagram
1. LLM Basics	Understand model behavior, context, tokens, hallucination, and cost	LLM Basics	Explain what happens in one model API call
2. Prompt Engineering	Design stable, reusable, testable prompts	Prompt Engineering	Reusable prompt template
3. RAG	Build document Q&A and knowledge-base systems	RAG	Personal knowledge base assistant
4. Agent	Let models call tools and execute workflows	Agent	Weekly report assistant or coding assistant
5. MCP	Connect tools, data sources, and AI clients through a shared protocol	MCP	Minimal local MCP server
6. Evaluation	Evaluate quality, stability, safety, latency, and cost	Evaluation	Eval dataset and scoring script
7. Production	Deploy with logging, permissions, monitoring, fallback, and cost control	Production	Production-ready AI app
8. AI Coding	Use AI coding tools effectively	AI Coding	Personal AI coding workflow
9. Interview	Prepare AI application engineering interviews	Interview	Clear project story

Practical Resources

Type	Resource	Useful For
End-to-end tutorial	Personal Knowledge Base Assistant	Build your first RAG project
Web tutorial	English Web Tutorial	Read online
Launch checklist	RAG Production Checklist	Moving a RAG demo toward production
Safety checklist	Agent Safety Checklist	Designing safer tool calling and Agents
Prompt checklist	Prompt Review Checklist	Making prompts more stable and testable
Failure cases	RAG Bad Cases	Understanding common RAG failures
Architecture template	AI App Architecture Template	Writing project README, design docs, and interview notes
Eval template	Eval Dataset Template	Creating AI app evaluation datasets

Minimal Code Examples

Example	Covers	Entry
Minimal RAG	Chunking, retrieval, context-grounded answers	examples/minimal-rag
Tool Calling	Tool schema, function call, result injection	examples/tool-calling
Minimal MCP Server	MCP tool definition and local file search	examples/minimal-mcp-server

Recommended Order

Start Here
  -> AI Application Foundations
  -> LLM Basics
  -> Prompt Engineering
  -> RAG
  -> Agent
  -> MCP
  -> Evaluation
  -> Production
  -> Tutorials / Checklists / Bad Cases
  -> Projects

Who This Is For

• Developers learning LLM application development systematically.
• Backend or full-stack engineers moving into AI application engineering.
• Builders working on RAG, Agents, AI assistants, knowledge bases, or automation workflows.
• Candidates preparing for AI application engineer, Agent engineer, or LLM engineer interviews.
• Product, operations, and data people who want to understand the AI app delivery lifecycle.

Not For

• People who only want to study model training or low-level ML algorithms.
• People who only collect links and do not build projects.
• People expecting to master all AI engineering topics in a few days.

Contributing

High-quality tutorials, project cases, bad cases, templates, and engineering lessons are welcome. Please read CONTRIBUTING.md first.

If this repository helps you, feel free to star it so more developers can find a practical path into AI application engineering.