when people evaluate LLM outputs, what’s hilarious now is that their prompts are highly influential to the outcome, unlike code where it’s much more objective

it’s kinda funny that everybody launched their cron jobs at 0:00 UTC

an internal tool that’s a wrapper of S3 with fewer features? why bother

Numbers on Mac doesn’t do a double-click to apply formula to all rows under this column…

i’m googling “beagle corgi mix” during this meeting

i don’t understand why my coworker says “i’ll send you the updated file” vs “let me open a PR and commit it to a team-owned repo”

planning things that are due in like 6 months…come on, let’s prioritize

when a meeting has too much crosstalk

having to lock in on a friday 😔

it’s an interesting tension when the manager thinks we’re not doing enough and an underling thinks we risk doing too much in too little time