RL me until I can contest program at the speed of thought
Interviewing is a mess… by Luna NovaI envy language models. Why can't they put me in an RL environment and make me amazing at contest programming? I'm so slow. Every line typed manually feels like a mistake. I can describe what I want to a language model and get it done faster.
They should reinforcement learn me with verifiable rewards me until correct code can be streamed from my brain at the speed of USB HID.
I already don't have episodic memory so we have the fresh env part down right, all we need is to ensure that if I go off task and start scrolling instead of programming that gets downweighted. How hard can it be?
I was part of the large Block layoff of '26, so I've been interviewing recently.
It's very strange going through coding challenges that involve manually typing things out when I rarely do so these days.
Can I take my interviews in Nix? Claude still has questionable enough Nix taste that I do that myself…
I wouldn't be in tech if I didn't like automating things and avoiding busywork, so interviews are more grating now than they ever have been in the past with the knowledge that: a machine can do this better.
Claude, invent vibe interviewing. Make no mistakes.
Cite as BibTeX
@online{coding-interviews-in-2026,
author = {Luna Nova},
title = {RL me until I can contest program at the speed of thought},
date = {2026-03-24},
year = {2026},
url = {https://lunnova.dev/articles/coding-interviews-in-2026/},
}