· 8 min read
Scaling coding agents past one machine
One MacBook capped me at two projects with coding agents and a full disk. Agents on a homelab VM and a PC, one client everywhere, took that to five.
What I learn along the way: AI, search and building for the web.
Written when there is something worth the length. Subscribe by RSS.
· 9 min read
How to boost your development speed with AI coding tools (without breaking everything)
Faster delivery with AI coding tools without shipping insecure code: foundations, prompts, security scanning, feature flags and MCP for React and TypeScript.
· 9 min read
Small-to-Medium language models you can run on your laptop — fully offline
Five 7B–13B language models that run offline on a laptop: Qwen2.5, Gemma 2, Mistral 7B, Xwin-LM and DeepSeek-R1 Distill, with sizes, hardware and licenses.
· 4 min read
CSS-in-JS is dead, long live CSS-in-JS
Runtime CSS-in-JS slows React apps down. Why teams are moving to zero-runtime options like Vanilla Extract and Tailwind, and what they gain.
· 3 min read
From Lexical Search to Semantic Search, and Their Role in RAG & Modern LLMs
How search moved from keyword matching to embeddings, and how retrieval-augmented generation pairs that retrieval with LLMs for context-aware answers.