Retraining & Rollout

Retraining as a decision rather than a schedule, champion/challenger, shadow, canary, A/B — and rollback and fallback, which every serving system needs before it needs a second model.

Retraining as a Decision
▶ lab

Retraining is a change to a running system with a cost, a risk and a precondition. Four questions decide whether it is due; "drift" is not one of them.

Q · The drift monitor fired, the model is six weeks old and the team is asking whether to retrain. How do you decide, and what makes the answer "no"?
Retraining Strategies
▶ lab

Scheduled, drift-triggered, performance-triggered, manual and continuous: each is right for a particular ratio of label speed to world speed, and continuous training has risks the others do not.

Q · Which retraining trigger fits this system, and what does continuous training cost that a scheduled retrain with a human gate does not?
Champion / Challenger
▶ lab

A candidate earns production by beating the incumbent on the same traffic under the same threshold policy, on more than one number. A better validation score is a nomination, not a promotion.

Q · A retrained model scores better offline than the one in production. What comparison justifies replacing the incumbent, and why is the validation delta not it?
Shadow Deployment
▶ lab

The candidate scores production inputs but controls nothing. It catches skew, latency and crashes before a user sees them — and it cannot measure business impact, because it never makes a decision.

Q · What can a model running in shadow tell you that offline evaluation cannot, and what can it never tell you?
Canary Rollout
▶ lab

Give the candidate 1% of decisions, then 5%, 25%, 100%, watching quality, latency, cost and errors at each step — with the honesty that a 30-day label makes a 30-day canary.

Q · How do you roll a model out so that a mistake hurts a few users instead of all of them, and what do you watch at each step when the real outcome will not arrive for a month?
A/B Testing Models

The only measurement of business impact is to let two models decide for two comparable populations and compare what happens — with stable assignment, guardrails, enough sample, and honesty about interference and about the users in the experiment.

Q · Why is a live experiment the only way to know whether a model improved the business, and what makes such an experiment invalid?
Rollback & Fallback
▶ lab

When the model is wrong, go back; when the model is gone, degrade. Rollback must restore the feature definition with the artifact or it reintroduces skew, and the previous model has to be warm.

Q · A model deployment is hurting the product. How do you go back in one step, and what does the product do while there is no model at all?