Start Here

This site is organized into series — sequences of posts that build on each other. There are a lot of them now, so this page is the map. Each entry links to the best first post; every post links onward from there.

Search, retrieval, and text

How machines find things in text — from the data structures up.

• Tries: Multi-Pattern Text Search — one data structure, six posts: scanning, autocomplete, compression, Spark, visualization.
• Entity Detection: Finding What Matters in Text — why finding names in text is harder than Ctrl+F, and how to score what you find.

Distributed data and approximation

Computing over more data than fits anywhere, and when a good estimate beats an exact answer.

• Mergeable Operations in Distributed Computation — the algebra that makes distributed computation composable.
• HyperLogLog: Counting Unique Items at Scale — building the classic cardinality sketch from scratch.
• Hulu Data Platform — a nine-post case study of a real event pipeline: DSLs, monitoring, graphs, and the reporting layer.

Machine learning and high dimensions

• Neural Nets from Scratch — from “simpler than you think” through mixture-of-experts.
• Exploring High-Dimensional Data — embeddings, projections, clusters, and a 3D explorer to poke at them.

Data stories and analysis

• NYC Taxis and Tipping — 88 million taxi rides, every bar in New York, and one question about tipping.
• Python for Fixed-Income Risk Analysis — yield curves, PCA, and volatility modeling with real Treasury data.

Building things

• Small Experiments — canoe trip maps, a music-amp control panel, an AI-extended puzzle curriculum, a drawing robot.
• Learn: Daily Thinking Puzzles — not posts but puzzles: self-paced tracks of five-minute problems in functional thinking, Bayesian statistics, cognitive science, and machine learning.

Standalone

• On AI and Authorship — how AI is (and isn’t) used in the writing here.