Showcasing my work and passion.
NSIGHT AI BUDGETING
Predictive Financial Modelling & Anomaly Detection
PYTHON • FASTAPI • STREAMLIT • SCIKIT-LEARN
ML-Powered Expense Classification
Manual reconciliation is slow and error-prone. I engineered a supervised learning model using Scikit-learn that predicts expense categories with 85%+ confidence based on historical vendor data, flagging low-confidence entries for human review.​​​
Time-Series Budget Forecasting
Static budgets fail to account for seasonality. I built a predictive engine that analyzes historical spend patterns to project future burn rates. Stakeholders can visualize variance before it happens, allowing for proactive capital allocation.​​​
Real-Time Variance Analysis
Stakeholders needed instant visibility into burn rates. I built a comparison engine that tracks YTD actuals against predicted budgets in real-time. The system calculates variance percentages and visually flags overspend risks (red) before they impact the bottom line.
Automated Anomaly Detection
To ensure data integrity, I implemented a statistical monitoring system that scans for irregularities. Using Z-score analysis, the system automatically flags transactions that deviate significantly from historical patterns, creating an early warning system for potential fraud or data entry errors.
