31
SQL patterns I use to catch transaction fraud
The actual queries I run when I'm hunting fraud in transaction data. Velocity, impossible distances, suspicious amounts, merchant clusters, off-hours buys, and window functions.
32
33
Does Reddit have an astroturfing problem? What the data suggests
In 51,129 comments from six knife subreddits, the 5% most brand-heavy accounts wrote 11.3% of the brand mentions in buying threads, against 7.9% expected by chance. For one brand it is 31% against 8%. The full Reddit histories of those accounts are 4.5 years old and spread over 66 subreddits, the same as a comparison…
34
Why DuckDB is my first choice for data processing
Why DuckDB has become my go-to tool for data processing, offering simplicity, speed, and powerful features.
35
My ridiculously robust photo management system (Immich edition)
How I added Immich to my already robust photo management system to make it ridiculously robust.
36
Text classification with Python 3.14's ZSTD module
Python 3.14 added Zstandard to stdlib; its incremental API finally makes classify-by-compression practical.
37
Using PostgreSQL as a Dead Letter Queue for Event-Driven Systems
Learn how to use PostgreSQL as a Dead Letter Queue (DLQ) for event-driven systems. Case study from Wayfair showing how to handle failures in Kafka consumers with PostgreSQL.
38
SQLite as a Document Database (2020)
However recently it added a killer feature: generated columns . (This was added in 3.31.0, released 2020-01-22.)
39
I analysed 20 years of my chats
Importing 1.2 million messages from Telegram, VK, Instagram, Facebook, and Twitter into a structured Obsidian vault with local LLM inference - to measure how my friendships work with decay curves, vocabulary shifts, directional sentiment, and conflict rates.
40
Detecting LLM-Generated Texts with “Classical” Machine Learning
.admonition { margin: .75em 0; padding: .6rem; overflow: hidden; font-size: 12px; page-break-inside: avoid; border-left: .3rem solid #42b983; border-radius: .3rem; box-shadow: 0 0.1re
41
Modern SQLite: Features You Didn't Know It Had
SQLite has evolved far beyond a simple embedded database. Explore modern features like JSON, FTS5, window functions, strict tables, and more.
42
Fabrice Bellard's TS Zip (2024)
Results and speed for other programs on enwik8 and enwik9 are available at the Large Text Compression Benchmark .
43
P99 0 ms* autocomplete for 240M domain names
How Wirewiki's domain-name autocomplete renders suggestions next frame for 99% of keystrokes.
44
MinIO Is Dead, Long Live MinIO
MinIO’s repo is officially archived and abandoned. And how AI Agents helped bring it back from the dead. This post explains how a community fork restores the admin console and ships binaries via CI/CD pipeline.
45
Language models for text classification: From bag-of-words to Jev
A Visual Guide to RNNs, CNNs, Transformers, and Calibration, with Hands-On Experiments on Accuracy and Efficiency
46
How Compaction Works in Pi
Why compaction is needed for large language models and how Pi implements it
47
SQLite in Production: Lessons from Running a Store on a Single File
Run Rails on SQLite with WAL and Docker volumes while accounting for the concurrency failure that appears during blue-green deploys.
48
Full-Text Search with DuckDB
Pete Doherty is a NYC based software developer
49
5NF and Database Design
Author: Alexey Makhotkin squadette@gmail.com, (~3900 words) One of the goals of this publication is to deconstruct the traditional ways of teaching basic topics in relational databases. Previously we discussed the fourth normal form (4NF): “Historically, 4NF explanations are needlessly confusing”. Let’s discuss the…
50
Good CTE, Bad CTE
The planner treats CTEs very differently depending on how you write them. Here's what happens under the hood, version by version, through PostgreSQL 18.
51
Pandas Should Go Extinct
Pandas has taken us a long way, but it's time for a new bear
52
Data Compression Explained (2012)
Copyright (C) 2010-2012, Dell, Inc . You are permitted to copy and distribute material from this book provided (1) any material you distribute includes this license, (2) the material is not modified, and (3) you do not charge a fee or require any other considerations for copies or for any works that incorporate…
53
Replacing a 3 GB SQLite db with a 10 MB FST (finite state transducer) binary
Note for numberphiles: all numbers have been rounded to their first significant digit, because I’m a fan of Rob Eastaway’s “zequals” method of getting to the point when it comes to estimation. It’s much more valuable to walk away with the heuristic “some dude got a 300x memory reduction by swapping out a database he…
54
Deep dive into Turso, the “SQLite rewrite in Rust”
I love Rust and I love SQLite, so you can guess. Iwas pretty excited when I lerned that "SQLite was rewritten in Rust" What is SQLite, actually? 2 things: a
55
Planet Labs' open satellite feed
Benchmarks & Tips for Big Data, Hadoop, AWS, Google Cloud, PostgreSQL, Spark, Python & More...
56
What category theory teaches us about dataframes
Every dataframe library ships with hundreds of operations. pandas alone has over 200 methods on a DataFrame. Is pivot different from melt? Is apply different from map? What about transform, agg, applymap, pipe? Some of these seem like the same operation wearing different hats. Others seem genuinely distinct. Without a…
57
Your File System Is Already A Graph Database
Karpathy recently posted about using LLMs to build personal knowledge bases — collecting raw sources into a directory, having an LLM “compile” them into a wiki of interlinked markdown …
58
UUIDv7: Timestamps aren’t time-ordered and other oddities
When timestamps aren’t time-ordered and other oddities
59
The perils of UUID primary keys in SQLite
A blog mostly about Clojure programming
60
It's Not Just X. It's Y
Against the Quantification of Integrity When the measure of language becomes its target, it ceases to be good language. 💡Nerd Rating: 1/5. I discuss the origins of certain linguistic tics in LLMs and what it means for writing, student assessment, and thinking. "It's not x, it's y.