A Small Result Set Can Hide a Huge Query
Two PostgreSQL investigations show how small batches and zero-row inserts hide repeated work—and how to prove an improvement without claiming optimality.
Blog
How I Made Apple Notes Change Detection 60× Faster
A local Apple Notes sync dropped from 49 seconds to 825ms by replacing N+1 Apple Events with vectorized JXA. Benchmarks against AppleScript, SQLite, file stats, and hashing show why the fastest query was not the best design.
llama.cpp vs MLX/oMLX: Architecture Benchmark on Apple Silicon
Head-to-head benchmark of llama.cpp vs Apple MLX for local LLM inference on M5 Max. Architecture comparison showing why fewer layers = faster inference — llama.cpp is 21-59% faster across all payload sizes.
Why Cargo Ate 45GB Overnight (And How I Fixed It)
Rust's incremental compilation cache never garbage collects. With multiple AI agent sessions running cargo across git worktrees, target/ ballooned to 45GB. Here's the root cause analysis, the empirical fixes, and what I rejected after adversarial auditing.