A space for curious questions
Pull a lever.
See what changes.
The best way to understand a project is to explore it.
Change a budget, compare a player, question a model.
The evidence is right here with you.
Choose your question
Your own point of view.
Explore the output.
Follow the evidence into a case study.
Preparing the interactive output…
Benchmark controls recalculate recorded evidence in your browser. Public apps load when you choose.
The details matter
A result is only as useful
as its context.
Every project carries its data source, the method behind it, and the limits of what it can tell you.
01 Recorded benchmarks
Fraud, churn, sentiment and click-through ranking use exported results from reviewed repository revisions. Changing a control explores those results. Saved model examples keep their original inputs and recorded scores.
02 Illustrative datasets
Booking patterns use 1,000 deterministic synthetic records. Talent Radar uses fictional companies and jobs. Both are labelled in the workspace so you can explore the approach with the right context.
03 Original experiences
Watch Atlas and Country Memory Map open the original public applications on request. PlayerPulse uses a saved player archive; the comparison shows missing coverage rather than filling gaps with invented values.