Engineering leader and hands-on polyglot engineer working across backend, platform, and AI tooling.https://kiran-capoor94.github.io/https://kiran-capoor94.github.io/blog/forced-breaks-in-neovim/https://kiran-capoor94.github.io/blog/forced-breaks-in-neovim/The forced-break Pomodoro system in Wand.nvim, and an undo-corrupting file write an AI code review caught inside it.Mon, 06 Jul 2026 00:00:00 GMThttps://kiran-capoor94.github.io/blog/neither-home-nor-riverbank/https://kiran-capoor94.github.io/blog/neither-home-nor-riverbank/On staffing/consulting work, an old Hindi idiom about a washerman's dog, and the specific moment you realize nobody in the chain actually owns your career.Mon, 06 Jul 2026 00:00:00 GMThttps://kiran-capoor94.github.io/blog/neovim-under-100ms/https://kiran-capoor94.github.io/blog/neovim-under-100ms/Wand.nvim's boot-time target comes from making lazy-loading the default everywhere, not from a single clever optimization.Mon, 06 Jul 2026 00:00:00 GMThttps://kiran-capoor94.github.io/blog/triage-without-an-llm/https://kiran-capoor94.github.io/blog/triage-without-an-llm/Wizard's work-triage scorer is a weighted sum, not a model call. The LLM only writes the sentence explaining the ranking, it doesn't produce it.Mon, 06 Jul 2026 00:00:00 GMThttps://kiran-capoor94.github.io/blog/unshipping-toon/https://kiran-capoor94.github.io/blog/unshipping-toon/A CSV-based compact encoding for AI agent memory saved ~40% tokens on paper. It broke local LLM parsing in practice, so I reverted it.Mon, 06 Jul 2026 00:00:00 GMT