Arun Sharma / ReadingThe RAG Interview

A complete study edition

The RAG Interview

Retrieval, Ranking, Grounding,
and System Design.

From first principles to production trade-offs. Explore the entire Notion study collection, organized for reading, practice, and reference.

41 chapters12 parts7 appendices

Start reading Choose a reading route

Source metadata: Hao Hoang · AI Interview Prep · First edition, 2026.
About this edition and source rights

The curriculum

Build the system.
Understand every trade-off.

Twelve parts, from framing the question to designing systems at scale. Open a part to explore its chapters.

01The RAG Interview LandscapeChapters 1–3
  1. 1: What a RAG Interview Actually Tests
  2. 2: The Competing Answers to the Same Problem
  3. 3: A Repeatable Framework for Any Retrieval-Augmented Generation (RAG) Design Question
02The Generator SideChapters 4–8
  1. 4: What the LLM Brings and What It Breaks
  2. 5: Data, Privacy, and the Legal Surface
  3. 6: Scaling Laws and the Economics of Retrieval vs Parameters
  4. 7: In-Context Learning - The Mechanism RAG Rides On
  5. 8: Reading the Machine: Circuits, Induction Heads, and Attribution
03Prompting and Context ConstructionChapters 9–11
  1. 9: Prompting for Retrieval-Augmented Systems
  2. 10: Prompt Sensitivity
  3. 11: Abstention and Calibration
04Representing What You RetrieveChapters 12–14
  1. 12: Text Representation
  2. 13: Chunking and Granularity
  3. 14: Beyond Plain Text - Tables, Layout, Documents
05Indexing and Vector SearchChapters 15–17
  1. 15: Approximate Nearest Neighbor Search
  2. 16: Compression and Index Economics
  3. 17: Operating a Vector Store
06Retrieval and RankingChapters 18–23
  1. 18: Classical Information Retrieval (IR) You Are Expected to Know
  2. 19: Learning to Rank
  3. 20. Dense Retrieval
  4. 21: Learned Sparse and Multi-Vector Retrieval
  5. 22: Reranking
  6. 23: Generative Retrieval
07Query Understanding and Control FlowChapters 24–26
  1. 24: Query Reformulation
  2. 25: Routing and Adaptive Retrieval
  3. 26: Iterative, Recursive, and Agentic Retrieval
08Training the RAG SystemChapters 27–29
  1. 27: Fine-tuning the Generator
  2. 28: Training the Retriever
  3. 29: Bootstrapping Training Data
09Generation and Context AssemblyChapters 30–31
  1. 30: Lost in the Middle
  2. 31: Attribution and Citation
10EvaluationChapters 32–34
  1. 32: Evaluating Retrieval
  2. 33: Evaluating Generation
  3. 34: Evaluating the System
11Trust, Credibility, and Adversarial RobustnessChapters 35–36
  1. 35: Source Credibility
  2. 36: Provenance and Adversarial Robustness
12Scaling, Advanced Variants, and Design DrillsChapters 37–41
  1. 37: Latency, Cost, and Systems
  2. 38: Distributed and Federated RAG
  3. 39: Multimodal RAG
  4. 40: Graph RAG
  5. 41: End-to-End Design Drills

Reference & practice

Keep the essentials close.

01 / ReferenceFormula Sheet02 / RehearseQuestion Bank03 / ReviewDesign Checklists04 / DecodeNotation and Symbols

Complete contents →Source index →

The book’s starting model

Knowledge in weights.
Evidence in a datastore.

Retrievable knowledge and model parameters play different roles. This original diagram from Chapter 1 introduces the two paths.

Chapter 1: the semi-parametric frame
Chapter 1: the semi-parametric frame · Open full size (new tab)
Diagram description and source
flowchart TD
    Q[query x]
    D[D: non-parametric datastore. Capacity grows with stored rows. Write one embedded chunk and row. Reversible in seconds.]
    W[[θ: parametric weights. Capacity fixed before training. Write with a gradient pass over every parameter. Entangled and measured in GPU-hours.]]
    R[RET returns c_1 through c_k]
    G[GEN reads x and c_1 through c_k]
    Y[answer y]
    Q -->|retrieve| D
    Q -->|condition| W
    D --> R
    R --> G
    W --> G
    G --> Y
    style D stroke-dasharray: 5 5

Take it with you

Read here. Study anywhere.

Prefer your own workspace? Download the complete Notion import, or browse the portable Markdown source on GitHub.

Download Notion import ↓View on GitHub ↗

ZIP · 1.5 MB · All 60 book pages, plus import instructions and attribution.
Import instructions · SHA-256 checksum

Contents

Search the book

Search all 60 pages. Try “chunking”, “latency”, or “evaluation”.