Speculative Decoding in vLLM: Draft, Verify, and Cut Latency Without Losing the Distribution
Large language model inference has an awkward performance profile: the GPU does enormous math, then waits. Every token requires a
Author: teliaz
Chesterton’s Fence and the Ethics of Deleting Legacy Code
Every codebase has one. A function nobody wants to touch, guarded by a check so strange it looks like vandalism:
Continue readingChesterton’s Fence and the Ethics of Deleting Legacy Code
Row Stores vs Column Stores: Why Analytics and OLTP Are Opposites
Every data engineer eventually has the same moment: the dashboard query that was instant on a million rows takes ninety
Continue readingRow Stores vs Column Stores: Why Analytics and OLTP Are Opposites
PostgreSQL vs MySQL in 2026: A Practitioner’s Comparison
PostgreSQL versus MySQL is one of the oldest rivalries in software, and like most old rivalries it is mostly narrated
Continue readingPostgreSQL vs MySQL in 2026: A Practitioner’s Comparison
Processes vs Threads vs Coroutines: The Units of Concurrency
“It’s slow” is the least useful bug report in software. “It’s slow when 400 users hit the checkout endpoint simultaneously,
Continue readingProcesses vs Threads vs Coroutines: The Units of Concurrency
Strings, Bytes, and Unicode: Why ‘café’ Breaks Your Code
Most developers treat strings as a solved problem: a string is text, text is easy, move on. Then one day
Continue readingStrings, Bytes, and Unicode: Why ‘café’ Breaks Your Code
How Hash Tables Actually Work: Collisions, Load Factors, and the Resize Dance
Every language you have ever used has a hash table hiding somewhere in it. Python calls it dict, Go calls
Continue readingHow Hash Tables Actually Work: Collisions, Load Factors, and the Resize Dance
SLOs You Can Actually Alert On: Error Budgets and Burn Rates in Prometheus
Every on-call rotation has a story about the alert that fires at 3 a.m. for something that isn’t wrong. CPU
Continue readingSLOs You Can Actually Alert On: Error Budgets and Burn Rates in Prometheus
Killing Flaky Tests: Reproduce, Quarantine, Fix
A test that fails every time is annoying. A test that fails some of the time is corrosive. The always-failing
Continue readingKilling Flaky Tests: Reproduce, Quarantine, Fix
Go 1.27: Generic Methods, JSON v2 by Default, and a Runtime That Finds Its Own Leaks
Six months of Go development lands at once twice a year, and the Go 1.27 release (August 2026) is one
Continue readingGo 1.27: Generic Methods, JSON v2 by Default, and a Runtime That Finds Its Own Leaks