This project demonstrates how AI and DevOps can be integrated to automate risk scoring of financial audit logs using unsupervised learning. Built and executed entirely on Azure DevOps, this project highlights real-world MLOps practices for internal audit modernization.
- Applied Isolation Forest to detect anomalies in simulated audit logs
- Built a fully automated CI/CD pipeline in Azure DevOps for data simulation, model scoring, and result logging
- Showcased reproducible and modular Python + YAML-based AI workflow
- Demonstrates transparency and automation for audit and compliance workflows
You’ll find:
- All code:
simulate_audit_data.py,run_pipeline.py, model artifacts - Azure pipeline YAML (
azure-pipelines.yml) - Logs and artifact history
- Public pipeline execution dashboard
| Tool | Role |
|---|---|
| Python | ML modeling and simulation logic |
| Scikit-learn | Unsupervised model (Isolation Forest) |
| Azure DevOps | CI/CD and orchestration |
| YAML | Pipeline-as-Code configuration |
| Timestamp | User | Action | Entity | Risk Score | Anomaly |
|---|---|---|---|---|---|
| 2023-02-09 23:00:00 | admin | delete | accounts | 0.0491 | True |
| 2023-01-31 12:00:00 | guest | modify | payroll | 0.0291 | True |
| 2023-01-22 01:00:00 | admin | access | inventory | 0.0259 | True |
| 2023-01-12 01:00:00 | admin | access | payroll | 0.0170 | True |
| 2023-01-19 02:00:00 | guest | modify | billing | 0.0144 | True |
This project is ideal for:
- Internal audit teams seeking automation
- Risk analysts evaluating log-based anomalies
- Engineers implementing CI/CD for ML pipelines