muneebullah muneeb

I've built end-to-end machine learning solutions for customer churn prediction, fraud detection, and time-series anomaly detection using XGBoost and LSTM Autoencoders.

For a Telco churn project, I used XGBoost and SHAP to identify key churn drivers, addressed class imbalance, and evaluated results with Precision, Recall, F1-Score, and ROC-AUC. I also built FraudSight, a binary classification system for detecting fraudulent financial transactions through preprocessing, scaling, and model optimization.

In time-series work, I designed an unsupervised LSTM Autoencoder that detects anomalies from reconstruction error and created temporal pipelines with sequence generation and normalization.

I'm currently pursuing a BS in Data Science at the University of the Punjab and have published over 20 machine learning notebooks on Kaggle. I enjoy turning complex data into actionable business insights through feature engineering, exploratory analysis, and clear visualizations.