Vendor Invoice Intelligence System

Freight Cost Prediction + Invoice Risk Flagging (End-to-End ML Project)

Build a production-ready Machine Learning system that solves a real finance problem. This is a business-driven, dual-model ML system with deployment and UI. If you want a project that makes recruiters stop scrolling your resume, this is it.

Why This Project Is Powerful for Your Career

1. It shows business understanding
2. Most ML portfolios show model training.
3. This project shows business objectives, financial impact, audit logic, and risk detection.

You are demonstrating capability in:

• Cost prediction (MAE, RMSE, R²)

• Risk classification (Precision, Recall, F1-score)

• Feature importance analysis

• Model comparison and selection

Few candidates show both in a single cohesive system.

1. It includes Python + SQL + ML integration
2. The project pulls data from a relational SQLite database and performs invoice-level aggregation.

This mirrors real production pipelines where data lives in structured databases not CSV files.

It is deployment-read, includeing:

• Saved trained models (.pkl files)

• Scaler objects

• Inference scripts

• Streamlit web application

• Real-time prediction interface