RESEARCH & NOTES

Insights on Machine Learning,
Algorithms & Data

Practical technical writing from the work of building production models that turn existing business data into better decisions.

CURRENT FLAGSHIP PIECE
How We Build Models

Our data-first philosophy: start with the datasets your company already generates, engineer the right signals, choose and tune algorithms for the actual decisions your teams make, and ship models that improve lead prioritization, retention, and forecasting.

Read the full piece →

Topics We're Covering

Feature Engineering for Tabular Business Data
What actually predicts in lead and customer datasets vs. what looks good in notebooks.
Data Collection That Improves Models
Labeling strategies, instrumentation decisions, and avoiding the "more data" trap.
Evaluating Models for Real Decisions
Beyond AUC: calibration, lift at operating points, expected value, and drift monitoring.
Why Forecasting Models Degrade (and How to Build Ones That Last)
Seasonality, concept drift, and building forecasts your planners can actually use.
Churn Prediction: Finding Actionable Drivers
Not just "who will leave" — which behaviors and interventions actually change the outcome.
Production ML Without Massive New Infrastructure
Fitting high-leverage models to the data and systems companies already run on.

We're publishing new technical notes regularly. In the meantime, explore our models or reach out directly.

Read How We Build Models Get in touch
Also view as /blog