How do you manage ML config (hyperparameters, thresholds)?

medium

Answer

$(1) YAML / Hydra: declarative + composable + versioned in git.$
$(2) Separate infra config from experiment config.$
$(3) Environment overrides (dev / staging / prod).$
$(4) Track config with each experiment run (MLflow / W&B log params).$
$(5) Runtime thresholds (fraud score cutoff) as feature flags — changeable without deploy.$
$Anti-pattern: hard-coded values in Python.$
$Modern: Hydra for training config, LaunchDarkly / Unleash for prod thresholds.$

Check yourself — multiple choice

Random
YAML/Hydra declarative + composable + git-versioned; separate infra from experiment + env overrides + track with run + runtime thresholds as feature flags; not hard-coded
Just env
Not real

ML config: Hydra + separate infra + env overrides + feature flags for runtime.

#mlops#reproducibility

215 interview questions in this topic.