The Spreadsheet That Beat Our Churn Model
We spent six weeks on a gradient-boosted churn predictor. A three-column spreadsheet from the support lead did better.
Ethan Cole
@ethan.aiML engineer in Seattle building practical AI products. Skeptical of hype, excited by results.
9 articles350 clapsJoined May 1
Articles
Stop Sending Your CEO the Average. Send the Shape Instead
Why a single mean hides the story in most business data, and three simple alternatives that lead to better decisions.
Your First AI Feature Should Be Boring
The AI features that survive contact with real users are rarely the flashy ones. Here is why I start with the dull jobs.
Evals Before Vibes: How We Actually Test LLM Features
Clicking around in a playground is not testing. A small, honest evaluation set will tell you more than a week of demos.
7 Questions I Ask Before Training Any Model
Most failed ML projects fail before the first line of training code. These questions catch the problems early.
How We Cut Our LLM Bill by 70% Without Users Noticing
Large model costs creep up quietly. A few unglamorous changes brought ours down sharply while quality held steady.
The Dashboard Nobody Opened: What 40 Unused Charts Taught Me
I audited the view logs of a dashboard I was proud of. The results were humbling, and they changed how I build reports.
Seven Questions I Ask Before Touching Any New Dataset
A checklist born from projects that went wrong, so that the next analysis starts from understanding instead of assumptions.
A 30-Minute Eval Setup That Catches Most Prompt Regressions
You don't need a fancy platform to test prompt changes. A list of real examples and a tiny script got me most of the way there.
