Sourabh Mathur

About Sourabh Mathur

I turn data into decisions and build AI systems that actually work in…

Experience

1. Data Scientist

   Wipro

   Sep 2024 — Present · Bengaluru, IN

   Architected and deployed an end-to-end employee attrition prediction system in Microsoft Fabric, forecasting 6-month churn using 2 years of historical data and achieving 86% accuracy with XGBoost, surpassing Random Forest and CatBoost baselines.Conducted SHAP and feature importance analysis to identify key attrition risk drivers, delivering actionable insights that enabled HR teams to implement targeted retention strategies and reduce employee turnover.Developed a predictive model for candidate offer acceptance probability, improving recruitment efficiency by helping talent acquisition teams prioritize high-likelihood candidates and optimize outreach strategies.Engineered critical features for both attrition and offer acceptance models through advanced feature engineering techniques, significantly enhancing model robustness, interpretability, and predictive power.Implemented production-grade MLOps infrastructure using MLflow and Grafana for real-time model monitoring, including automated alerts for performance degradation, data drift detection, and model retraining triggers.Established comprehensive metrics tracking framework monitoring accuracy, precision, recall, F1-score, and drift metrics to ensure sustained model reliability, compliance with SLAs, and continuous performance optimization.Delivered fully automated ML workflows encompassing data ingestion, preprocessing, feature engineering, model training, deployment, and monitoring—enabling scalable, repeatable ML operations across the organization