Building a Native GTK4 PostgreSQL Client Because I Was Tired of Electron
Why I built a native GTK4 PostgreSQL client instead of another Electron app — schema designer, AI analytics, and a Python/Perl hook system included.
Aug 4, 2026
5 min
The Rise of AI Trading Among Retail Investors
Robinhood just let customers connect Claude and ChatGPT straight to a trading account. Zerodha, Alpaca, and a wave of no-code platforms have made AI t...
The Alpine Mirage: How Upgrading Python Broke My Build and Led to a Truer Security Posture
A deep dive into how upgrading a Python Docker image to Alpine 3.15 caused cascading build failures across C/Rust dependencies—and why switching to De...
Why Cosine Similarity Fails to Catch Confusable MCP Tools
Cosine similarity fails to catch confusable MCP tools. Here's the schema-substitutability approach that actually worked, packaged as an open-source li...
Building a Zero-Trust Internal API on AWS
A DynamoDB table, a read-only Lambda, an API Gateway locked to IAM auth, and an EC2 client with no SSH key: a full zero-trust internal API on AWS, wit...
Arrays Explained from Memory to Big O (With Real Benchmarks)
A common array bug that silently deletes the wrong items, and what it reveals: how arrays actually work in memory, why insert/delete cost what they co...
Linear Algebra Essentials: The Math Every Model in This Series Has Been Hiding
Every prediction this series has made was a dot product wearing a library's clothing. Here's the linear algebra underneath, verified against sklearn's...
Statistics Essentials for Machine Learning: Mean, Variance, Sampling, and Significance
Day 4 left a real question open: was a survival gap signal or noise? Today's statistics tools actually answer that, instead of just shrugging at it.
Data Quality: Why a 96% Accurate Model Can Still Be Completely Useless
One column is 77% empty. One model is 96% accurate and catches nothing. Here's how to check your data before you trust what your model tells you.