Build an End-to-End ML Pipeline for Loan-Default Prediction
Overview
What this challenge is about.
You receive 24 months of historical application + outcome data (about 380,000 rows). Build a pipeline using a workflow orchestrator (Prefect, Kedro, or a simple Makefile chain) that does: raw-data validation, feature engineering, time-aware train/test split, model training (gradient boosting baseline), evaluation against a defined business KPI (approval rate at fixed default rate), and model packaging. Add at least 6 unit tests on the feature layer. The pipeline must run end-to-end in under 20 minutes on a single 4-vCPU machine.
The Brief
What you'll do, and what you'll demonstrate.
Build a reproducible, tested, end-to-end ML pipeline for loan-default risk that engineering can productionize.
Earning criteria — what you'll demonstrate
- Design and implement a reproducible end-to-end ML pipeline
- Apply pipeline-style unit testing to feature-engineering code
- Tie model evaluation to a real business KPI, not just AUROC
- Hand off ML code in a state engineering can productionize
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.
Machine Learning Engineer
End-to-end, tested ML pipelines hand-off-able to engineering are the headline portfolio piece for any MLE role in fintech or other regulated industries.
This challenge sharpens
- ml-pipelines
- pipeline-testing
- reproducibility
MLOps Engineer
Reproducible pipelines, packaged artifacts, and a runnable scoring entrypoint are exactly what MLOps engineers expect when they take ownership of a model.
This challenge sharpens
- ml-pipelines
- reproducibility
- python
Data Engineer
Feature-layer testing and deterministic pipelines bridge directly into data-engineering work on feature stores and ETL orchestration.
This challenge sharpens
- feature-engineering
- pipeline-testing
- ml-pipelines