chiphuyen/aie-book

★ 15,659Jupyter NotebookAudience · developerComplexity · 2/5Setup · easy

Mindmap

mindmap
  root((aie-book))
    What It Covers
      Foundation model apps
      RAG and fine-tuning
      Production AI systems
    Key Topics
      Hallucination reduction
      AI evaluation
      Faster cheaper inference
    Companion Materials
      Chapter summaries
      Prompt examples
      Jupyter notebooks
    Audience
      AI engineers
      ML engineers
      Technical PMs

mindmap root((aie-book)) What It Covers Foundation model apps RAG and fine-tuning Production AI systems Key Topics Hallucination reduction AI evaluation Faster cheaper inference Companion Materials Chapter summaries Prompt examples Jupyter notebooks Audience AI engineers ML engineers Technical PMs

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Use chapter summaries and study notes as a guide for engineers learning how to take an AI model from demo to production application.

USE CASE 2

Learn when to use RAG versus fine-tuning for a specific AI product use case, using the book's decision frameworks.

USE CASE 3

Analyze your ChatGPT or Claude conversation history with the included Jupyter notebook to generate usage heatmaps.

Tech stack

PythonJupyter Notebook

Getting it running

Difficulty · easy Time to first run · 5min

In plain English

This repository accompanies the book "AI Engineering" by Chip Huyen, published in 2025. The book covers how to build real-world applications using foundation models, a category that includes large language models (LLMs, the kind that power chatbots) and large multimodal models (models that can handle both text and images). The repository provides supplementary materials: chapter summaries, study notes, prompt engineering examples, case studies, an appendix, and a list of AI engineering resources. The book itself is not a coding tutorial with lots of code snippets. Instead, it is a conceptual and practical guide for engineers and technical roles who want to understand the full process of going from an AI model to a working production application. It covers questions like: when and how to use techniques like RAG (Retrieval-Augmented Generation, enriching an AI's answers with relevant external documents), how to detect and reduce hallucinations (false or made-up outputs), when to fine-tune a model (customize it on your own data), how to evaluate AI outputs, and how to make AI systems faster and cheaper to run. The target audience includes AI engineers, machine learning engineers, data scientists, engineering managers, and technical product managers who want a framework for building serious AI products rather than just demos. The repository includes a Jupyter Notebook (an interactive coding document) for generating conversation heatmaps from ChatGPT and Claude chat histories. The full README is longer than what was provided.

Copy-paste prompts

Prompt 1

Based on AI engineering best practices, design a RAG pipeline for a customer support chatbot that retrieves from a company knowledge base and reduces hallucinations.

Prompt 2

I'm building a production AI feature. Walk me through the decision: should I use RAG, fine-tuning, or a combination, and what factors should drive that choice?

Prompt 3

Write an LLM evaluation framework that scores responses on relevance, factual accuracy, and tone for a production chatbot, using LLM-as-a-Judge.

Prompt 4

How do I make my LLM application cheaper and faster to run in production without sacrificing output quality?

Prompt 5

Explain the full engineering loop for detecting and reducing hallucinations in a deployed LLM application.

Open on GitHub → Explain another repo

← chiphuyen on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.