Skip to main content

COMPEL Glossary / continuous-integration-ml

Continuous integration (ML)

Automated test and build of model code, data contracts, and training scripts on every change — extended from software CI with data-schema validation, model-schema validation, and lightweight training smoke tests.

What this means in practice

Prevents regressions in model code and training-data interfaces before they reach pipeline runs.

Synonyms

CI for ML , ML CI , ML continuous integration

See also

  • Continuous delivery (ML) — Automated, governed promotion of models through lifecycle stages — development, staging, production — with gated checkpoints (evaluation thresholds, bias checks, cost thresholds, human approval where required).
  • Pipeline — An automated execution graph connecting data ingestion, feature engineering, training, evaluation, and deployment stages — parameterized, versioned, and re-runnable.
  • Data leakage — Information from the test or validation set inadvertently entering training — through preprocessing, feature engineering, target encoding, or time-ordered splits — inflating offline metrics and producing over-optimistic ship decisions.
  • Experiment tracking — The infrastructure and practice of recording artifacts, metrics, parameters, environment, and lineage for every experiment run — enabling later reproduction, comparison across runs, and audit.