Sentinel

The problem

Sentinel is a fraud detection research/demo platform for anonymized credit card transaction data.

What I built

Sentinel is a fraud-scoring research project for anonymized transaction data. I built it to test the parts that are easy to hand-wave in ML demos: imbalanced evaluation, threshold choices, per-score explanations, and an analyst review loop that actually saves its state.

How it works

• Synthetic data is the default path so CI and local demos do not require a Kaggle account.
• Accuracy is intentionally not a primary metric. The dashboard and reports lead with PR-AUC, recall at fixed precision, cost-weighted loss, and threshold confusion matrices.
• Logistic regression, random forest, and isolation forest are implemented as sklearn model adapters. XGBoost, LightGBM, PyTorch autoencoder, stacked ensemble, and true SHAP remain tracked production gaps in docs/REQUIREMENTS_AUDIT.md.

Tradeoffs and limits

This project does not currently have a hosted demo. Run and verification instructions live in the repository when available.

What I learned

No real PII or real cardholder data belongs in this repository. Sentinel is designed for the public anonymized Kaggle ULB credit card fraud dataset or synthetic data generated by the project itself.