Why I built llm-queue: one local LLM (Ollama) behind an OpenAI-compatible HTTP API, so every process shares one serialized priority queue instead of thrashing the model.
Writing
Notes on React, Next.js, TypeScript, testing, and building frontends that hold up in production.
Browse by tag
Playwright can mock browser requests, but not the fetches your Next.js server makes during SSR. Here is how I record real API responses once and replay them deterministically on CI, with the backend turned off.
The story of building state-in-url: turning the browser URL bar into real, type-safe React state — the design decisions and trade-offs.
Define and use responsive breakpoints as variables in pure CSS — no preprocessor or JS — for cleaner, DRY media queries.
My experience with CodeRabbit, an AI code review tool that actually adds value to pull requests — what it catches and where it helps.
Automating npm package releases with GitHub Actions: versioning, changelogs, and publishing to npm from CI without manual steps.
Practical use cases for React's useInsertionEffect hook — what it's for, how it differs from useEffect and useLayoutEffect, and when to use it.
What actually separates clean code from bad code — practical heuristics and examples, beyond the usual book quotes.
TypeScript function overloading explained with a realistic, real-world example — when it beats union types and how to write overloads cleanly.
How to share serializable state between unrelated React components — including server and client components in Next.js — without prop drilling.