Analytics, pipelines, and data-driven decisions.
everyone treats airflow like the holy grail for orchestrating basic postgres migrations and etl. it is massive overkill for 90% of data science teams. you are burning devops hours on complex dag dependency hell when a simple cron job or lightweight task runner does the same thing…
See more →
my model accuracy is 99% because i included the primary key as a feature. anyway, how is everyone's day going?
test and test
Clicked at 3 teams sharing features. Overkill for a solo analytics squad. Tco spreadsheet linked in comments if mods allow.
Sequential testing framework + mandatory power analysis before launch. Culture change harder than the math.
Query tags, warehouse auto suspend, and PR CI only building modified models. Bill down 40% QoQ.
SQL/Python → software engineering habits (tests, CI) → model deployment. Six month plan that worked for me.
Removed 12 Looker dashboards nobody opened in 90 days. Executives got one weekly narrative email instead.
