Jonathan Putney — builder, vp/tech, husband, dad, gamer

▸ shipping the agentic ai platform that powers our llm features (rag, multi-judge eval, websocket streaming, governance)

▸ shipped ratchet v0.5.0 — a cluster-safe job scheduler for jakarta ee — and finished rebuilding its core on spring boot

▸ recalibrating weekly on what ai can and can't do in real work. every time i pin down the ceiling it moves, then it finds a new way to be confidently wrong about something i'd already trusted it with

we build a 21 cfr part 11 compliant lms for some of the most regulated buyers on the planet — pharma, life sciences, government, hospital systems, airports. i architect the platform, run engineering, lead compliance, and still ship production code most days. ~250k annual users. 40+ tenants. 6 eks clusters.

noverant laid me off in the 2008 recession. i stayed on good terms with the ceo anyway, and years later i came back for the room to grow and the autonomy to actually use it.

"player-coach" usually means the exec shows up to standup. here it means i argue a pricing position on an rfp in the morning and debug a schema migration after lunch.

a real one from this month: a migration that passed every test locally and would have taken down the production deploy. it dropped an index before creating its replacement, and innodb still needed that index to check a foreign key. we found it by running it against a throwaway mysql container instead of reasoning about it, which is the only reason we found it at all.

i've hired every technical person here for ten years. i still read the diffs, and i'm still the one who says no to the feature that would only ever help one client.