Day 1 Lightning Talks
summary
Quick checks for a bad job: shipping features over customer value, tests discouraged, and people doing untracked shadow work. Operational advice: return partial data instead of 500s and design for graceful failure; avoid microservices unless independent deployability or scale demand it. NLP claim: bag‑of‑words classifiers are cheap but miss negation and sarcasm; a simple movie gem demo can build a classifier. Language/tool demos: Amber on Crystal scaffolds fast apps. A Ruby maybe library (mombi maybe type) prevents nils and cuts exceptions. One team reports 100+ tickets shipped and 24,000 paying users with CI protecting deploys.
AI-written summary · about
