Reading Room
Books, papers, and side maps I keep returning to. A library that shapes how I think about AI, software, intelligence, and a few unrelated curiosities.
The shelves below cover the AI / software / strategy core. The Side maps are personal research projects on things I’ve gone deep on outside of AI — kept here so they live in one honest place rather than competing with the main work.
Six shelves
The library is organised the way I actually use it: by the question being answered, not by the discipline.
Artificial Intelligence
Books, papers, courses, and lectures on AI foundations, deep learning, language models, agents, and evaluation.
Software Systems
Resources on architecture, infrastructure, reliability, security, developer tools, and production engineering.
Intelligence & Cognition
Resources on biological intelligence, human learning, neuroscience, evolution, swarms, and decision-making.
Startups & Strategy
Resources on company building, distribution, positioning, sales, venture capital, and markets.
Trust, Risk & Regulation
Resources on compliance, auditability, software verification, AI governance, security, and institutional trust.
Writing & Communication
Resources on clear thinking, explanation, persuasion, storytelling, and public intellectual work.
Personal research projects
Two long-form maps I built outside the AI core, kept here because they are honest about being side projects. Educational only — not medical, legal, or financial advice.
The Peptide World
An educational map of peptide science, medicine, biohacking hype, safety risks, and the databases behind the field. Written with TGA and WADA caution. Personal research project — not medical advice.
The Scientific Process
A practical guide to science, evidence, uncertainty, bias, and how to form better beliefs in a noisy world. Evidence ladder, bias toolkit, and a belief framework. Personal research project, not AI work.
Foundational maps I build for myself
Longer than a blog post, narrower than an atlas. Field maps reduce overwhelm in domains I work in every day by organising them into one clear mental model.
The list
A starting set. Twelve resources chosen because each one teaches a model that keeps paying off in unrelated problems.
| Resource | Type | Difficulty | Why it matters | Related |
|---|---|---|---|---|
| Shelf 01 · Artificial Intelligence | ||||
| Artificial Intelligence: A Modern Approach | Book | Beginner | The canonical map of the AI field. Read once for vocabulary, again for structure. | Knowledge Bank → |
| Deep Learning | Book | Intermediate | Goodfellow, Bengio, Courville. The mechanics behind every modern model. | AI Atlas → |
| Attention Is All You Need | Paper | Advanced | The architecture that broke open the modern era. Read it slowly, then read it again. | Essay → |
| The Bitter Lesson | Essay | Beginner | Sutton, three pages. Why general methods that ride compute keep beating clever ones. | Field Manual → |
| Shelf 02 · Software Systems | ||||
| The Mythical Man-Month | Book | Beginner | Brooks on why adding people to a late project makes it later. Still true. | Labs → |
| Designing Data-Intensive Applications | Book | Intermediate | Kleppmann. The clearest map of storage, replication, and distributed tradeoffs. | Projects → |
| Shelf 03 · Intelligence & Cognition | ||||
| Gödel, Escher, Bach | Book | Advanced | Hofstadter on self-reference, pattern, and minds. Slow, strange, foundational. | Bio Atlas → |
| How Buildings Learn | Book | Intermediate | Brand on how systems evolve under use. Read it as a software book in disguise. | Blog → |
| Shelf 04 · Startups & Strategy | ||||
| Zero to One | Book | Beginner | Thiel. Why competition is for losers and monopoly is the prize worth building. | Money Map → |
| The Lean Startup | Book | Beginner | Ries on cycle time, hypothesis testing, and not falling in love with your demo. | Notes → |
| High Output Management | Book | Intermediate | Grove. The clearest operational handbook ever written for a company. | Notes → |
| Poor Charlie’s Almanack | Book | Intermediate | Munger on mental models, incentives, and avoiding stupidity. | Scientific Process → |
How I read
Three passes. The first is for understanding. The other two are for use.
Understand the main idea
Read for shape and argument. Resist note-taking. The goal is to grasp the whole before naming the parts.
Extract models and mechanisms
Slow second read. Write down the moving parts: assumptions, mechanisms, edge cases, names worth remembering.
Connect it to systems I am building
Map the ideas onto active work. If nothing connects, the book is interesting but not yet useful. Park it.
The shelves explain where the ideas come from. The atlases are what I do with them.
Source-backed maps of the AI field, built from the books and papers on this page.
Open atlases