Back to Portfolio
Supply Chain Analytics

Mining Supply Intelligence Platform

A 7-page analytics dashboard with 6 intelligence engines. Procurement risk, compliance monitoring, spend anomalies. Turns raw supply chain data into actionable intelligence before problems reach operations.

29 Automated Tests Seed-based Synthetic Data (seed=42) Portfolio Project
Procurement risk in mining is invisible until a supplier fails. By the time it hits your KPIs, it's too late. The warehouse is empty, the machine is idle, the shift is lost. This platform scores 500 suppliers across risk dimensions, monitors compliance across 6 weighted dimensions, and detects spend anomalies before they compound. Built for the procurement manager who needs to answer "which suppliers are about to cause me a problem?" before the problem happens.

Analytics Engines

🔌
Risk Engine
Random Forest

Supplier risk scoring: Low / Medium / High / Critical. Calibrated for 90% Australian supply base. Thresholds derived from operational failure patterns.

✅
Compliance Engine
6 Weighted Dimensions

Modern slavery compliance and ESG monitoring. Six weighted dimensions including labour practices, environmental reporting, governance transparency.

📈
Spend Analytics
Isolation Forest + 2-sigma

Anomaly detection across 50K transactions. Flags statistical outliers and pattern deviations. Weekend transaction threshold: $250K (mining is 24/7).

📄
KPI Engine
8 Headline Metrics

Executive KPI calculations across FY23-FY25. Australian fiscal year aligned. P&L-ready outputs for procurement leadership reporting.

🗣
NLQ Interface
Claude API (production) / Mock (demo)

Natural language queries against the full dataset. Ask "which suppliers are high-risk in WA?" and get a structured response. Demo uses MockNLQEngine.

📊
Visualizations
ChartFactory + Brand Theme

Custom chart factory with mining-major brand theme. Consistent visual language across 7 dashboard pages. Plotly-based interactive charts.

Platform Stats

500
Synthetic suppliers
50K
Transactions (FY23-25)
29
Automated tests
7
Dashboard pages

Technology Stack

Python 3.11 Streamlit Plotly Pandas scikit-learn pytest (29 tests) Claude API