04

Reading room

Library · Field maps · Method

The inputs
behind the work.

Books, papers and side projects organised by the question they help answer. This is the research layer behind the engineering, writing and atlases.

Core library

Twelve resources.
Models that keep paying off.

This is the complete curated list from the original site. Every item is retained here with the reason it matters and the part of the portfolio it informs.

01
Artificial Intelligence

Artificial Intelligence: A Modern Approach

BookBeginner

The canonical map of the field. Read once for vocabulary, then again for structure.

Knowledge Bank
02
Artificial Intelligence

Deep Learning

BookIntermediate

Goodfellow, Bengio and Courville on the mechanics behind modern neural networks.

Landmark papers
03
Artificial Intelligence

Attention Is All You Need

PaperAdvanced

The architecture paper that opened the modern era. Best read beside the systems consequences.

Read essay
04
Artificial Intelligence

The Bitter Lesson

EssayBeginner

Rich Sutton on why general methods that exploit computation keep winning.

Field Manual
05
Software Systems

The Mythical Man-Month

BookBeginner

Fred Brooks on communication cost, conceptual integrity and why late projects resist brute force.

Selected work
06
Software Systems

Designing Data-Intensive Applications

BookIntermediate

Martin Kleppmann's clear map of storage, replication and distributed trade-offs.

Auth case study
07
Intelligence and Cognition

Gödel, Escher, Bach

BookAdvanced

Douglas Hofstadter on self-reference, patterns and minds. Slow, strange and foundational.

Bio atlas
08
Intelligence and Cognition

How Buildings Learn

BookIntermediate

Stewart Brand on how systems adapt under use. It rewards being read as a software book.

Writing
09
Startups and Strategy

Zero to One

BookBeginner

Peter Thiel on differentiation, power laws and building something difficult to substitute.

Commercial map
10
Startups and Strategy

The Lean Startup

BookBeginner

Eric Ries on cycle time, hypothesis testing and learning before scaling.

Selected work
11
Startups and Strategy

High Output Management

BookIntermediate

Andy Grove's operational handbook for leverage, measurement and management systems.

Selected work
12
Startups and Strategy

Poor Charlie's Almanack

BookIntermediate

Charlie Munger on mental models, incentives and reducing avoidable error.

Scientific process
Six shelves

Organised by use,
not discipline.

Each shelf groups material that compounds around a recurring problem in the work.

01

Artificial Intelligence

Foundations, deep learning, language models, agents and evaluation.

Foundations · Frontier
02

Software Systems

Architecture, infrastructure, reliability, security and production engineering.

Build · Run
03

Intelligence and Cognition

Biological intelligence, learning, neuroscience, evolution and decision-making.

Mind · Pattern
04

Startups and Strategy

Company building, distribution, positioning, venture capital and markets.

Sell · Scale
05

Trust, Risk and Regulation

Auditability, verification, AI governance, security and institutional trust.

Verify · Govern
06

Writing and Communication

Clear thinking, explanation, persuasion, storytelling and public work.

Think · Explain
Field maps

Longer than an essay.
Narrower than an atlas.

Focused research projects kept here so they remain honest side paths rather than competing with the recruiter-facing portfolio.

Reading method

Three passes.
Understanding to use.

  1. 01

    Understand the main idea

    Read for the shape of the argument before naming every part.

  2. 02

    Extract models and mechanisms

    Capture assumptions, moving parts, edge cases and useful names.

  3. 03

    Connect it to active systems

    Map the idea onto work in progress. If nothing connects, park it until it does.

Research routes

From reading
to working systems.

The library is complete here. Continue into the writing archive for arguments, or the atlases for interactive maps of the ideas and infrastructure.

Browse all writing