When I onboard a new tech, I start them with environmental monitoring swabs at 08:00 before we open; it teaches aseptic flow, chain-of-custody, and clean documentation in one pass. For those coordinating benches, what early assignments have improved plate count accuracy or MIC reproducibility in your teams?
Right after “08:00 before we open,” I have new techs do a blind double-read on the same plates using a clear grid overlay, then we compare variance — it exposes edge/center bias and tightened our plate counts fast. If overlays feel fussy, a simple tally counter works too — do you track inter-reader agreement week to week?
When it’s 4:58pm Friday and someone wants a PNG “so the font looks perfect,” I ship the H1 as real text but inline a WOFF2 subset with only those glyphs (glyphhanger/fonttools) so it renders instantly and matches Figma; it’s about 20–30KB and avoids CLS. Small caveat: if the push for PNG is licensing from Figma, check web rights or swap to a licensed metric‑compatible face, else you’re just papering over risk. If you want a quick start, glyphhanger does the subset pass nicely: GitHub - filamentgroup/glyphhanger: Your web font utility belt. It can subset web fonts. It can find unicode-ranges for you automatically. It makes julienne fries..