DMML book guide -- what to read for each deck.

The uploaded slides cite several books rather than one single textbook. This guide maps each deck to the parts of those books that explain the same idea in deeper, cleaner language. Read the slide explanations first for exam shape; read these books for engineering depth.

Reading map by lecture

Lecture	Deck theme	Primary reading	Why
CS1	Representation, models, storage, serialization, OLTP/OLAP	Kleppmann + Reis/Housley	Explains storage, encoding, data models, transactions and analytical systems.
CS2	Data as asset/liability, platform, governance, life cycle	Reis/Housley + Reliable ML	Connects data lifecycle to business, governance and ML reliability.
L3	Architecture, warehouse/lake/lakehouse, mesh/fabric, Lambda/Kappa	Reis/Housley + Kimball + Kleppmann	Covers architecture patterns and why storage/processing choices differ.
L4	Data pipelines, services, message passing, modern data stack	Kleppmann + Densmore	Best explanation of dataflow, schema evolution, services and pipeline design.
L5	Infrastructure, storage systems, serving, DataOps, CT/CD	Tuulos + Reis/Housley + Reliable ML	Bridges data engineering infrastructure with model development and operations.

Fundamentals of Data Engineering

Joe Reis and Matt Housley -- Fundamentals of Data Engineering. This is the best spine for the course. The lectures use the same lifecycle framing: generation, storage, ingestion, transformation, serving, plus undercurrents like security, governance, DataOps, orchestration and architecture.

• Main idea: data engineering is not just ETL. It is the full lifecycle that turns generated data into reliable data products.
• Use for CS2: data management strategy, data platform components, lifecycle stages and governance.
• Use for L3: architecture choices: warehouse, lake, lakehouse, data mesh, data fabric, batch and streaming.
• Use for L5: storage systems, object storage, HDFS, orchestration, serving, reverse ETL and DataOps.
• Exam memory: always answer architecture questions using lifecycle + undercurrents: data moves through stages, and every stage needs security, governance, reliability and observability.

Designing Data-Intensive Applications

Martin Kleppmann -- Designing Data-Intensive Applications. This is the deeper systems book behind CS1 and L4. It explains the details that slides mention quickly: data models, storage, encoding, schema evolution, distributed consistency, services and stream processing.

• Data models: relational, document and graph models are not just storage choices; they express how the application thinks about relationships and queries.
• Encoding and evolution: serialization must survive rolling upgrades. Old code may read data written by new code, so compatibility matters.
• Storage: row/column tradeoffs connect to access pattern: transactional systems read/write records; analytical systems scan columns.
• Services and RPC: network calls fail differently from local function calls; hiding that difference creates brittle systems.
• Streams: logs and message brokers decouple producers/consumers and support replay, which is the foundation of CDC and Kappa-like designs.

Reliable Machine Learning

Cathy Chen, Niall Richard Murphy, Kranti Parisa, D. Sculley and Todd Underwood -- Reliable Machine Learning. This book explains why DMML exists for ML students: ML systems fail not only because model code is wrong, but because data, features, training pipelines, serving and monitoring drift out of alignment.

• Use for CS1/CS2: data sensitivity of ML pipelines and why missing regions, stale features or changed schemas can damage model behavior.
• Use for L5: CT/CD, model operations, observability, incident response and operational maturity.
• Core lesson: model quality is a property of the full sociotechnical system: data producers, pipelines, labels, features, code, deployment, monitoring and rollback.
• Exam memory: "pipeline succeeded" is not enough. Need data quality, distribution checks, feature freshness, model drift and output monitoring.

The Data Warehouse Toolkit

Ralph Kimball and Margy Ross -- The Data Warehouse Toolkit. This is the classic source for dimensional modeling: fact tables, dimension tables, star schema and data marts.

• Use for L3: warehouse schema, star vs snowflake and department-specific marts.
• Fact table: measurable business events such as orders, payments, shipments, clicks or support tickets.
• Dimension table: descriptive context: customer, product, time, geography, campaign, merchant.
• Exam memory: analytics schema is shaped by query pattern, not by perfect normalization.

Data Pipelines Pocket Reference

James Densmore -- Data Pipelines Pocket Reference. This is a practical companion for L4 and L5 pipeline questions.

• Use for L4: batch vs stream, ETL vs ELT, pipeline implementation, orchestration and failure handling.
• Use for L5: operationalizing pipelines with scheduling, retries, monitoring and data serving.
• Exam memory: pipeline design is source + destination + transformation + quality contract + failure behavior.

Effective Data Science Infrastructure

Ville Tuulos -- Effective Data Science Infrastructure. This helps with L5's data science/ML stack: warehouse, compute, scheduler, application architecture, versioning, feature engineering, model development and model ops.

• Use for L5: common ML infrastructure layers and why model development is only the top of a larger stack.
• Key idea: data scientists need freedom to iterate, but platform teams must provide reliable foundations so experiments can become production systems.
• Exam memory: reproducibility depends on versioning data, code, features, models and execution environment.

Fast reading order

1. Read DMML slides explained to understand the course flow.
2. Read Reis/Housley lifecycle chapters for the data-engineering backbone.
3. Read Kleppmann on data models, encoding/evolution, services and streams for CS1/L4 depth.
4. Read Kimball only for dimensional modeling and marts.
5. Read Reliable ML + Tuulos for ML pipeline reliability, CT/CD, model ops and infrastructure.

More in this vault

• References -- external links and tool map.
• Cheatsheet -- quick revision.
• Question Bank -- active recall.