Automate Retraining with a Drift-Triggered MLflow Pipeline
Overview
What this challenge is about.
Stand up the pipeline end to end with the team's existing stack (MLflow tracking + model registry, Airflow orchestration). Wire Evidently to compute weekly drift; when drift crosses threshold, kick off a retraining DAG. The DAG retrains on the latest 90-day window, evaluates against the production champion on a holdout, and promotes via MLflow registry stage transition only on a documented win condition. Include a manual-approval gate for healthtech compliance. Demonstrate one full cycle (manually-injected drift) and write a 4-page operations doc.
The Brief
What you'll do, and what you'll demonstrate.
Ship a drift-triggered retraining pipeline that auto-evaluates and promotes new models, with a compliance-friendly manual gate.
Earning criteria — what you'll demonstrate
- Implement an automated retraining pipeline with MLflow + Airflow
- Set drift-detection thresholds that fire on real shifts, not noise
- Design a promote-on-win gate with compliance-friendly approvals
- Document an audit-ready retraining process
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
MLOps Engineer
Drift-triggered retraining pipelines with MLflow + Airflow are the platform-MLOps work that regulated AI teams need to scale beyond manual retraining.
This challenge sharpens
- mlflow
- airflow
- automated-retraining
Machine Learning Engineer
MLEs increasingly own the retraining lifecycle end to end; this challenge gives a strong portfolio piece for that capability.
This challenge sharpens
- automated-retraining
- data-drift-detection
- model-registry
Data Engineer
Building the Airflow DAGs and data flows that underpin automated retraining is the data-engineering side of any ML platform team.
This challenge sharpens
- airflow
- data-drift-detection
- model-registry