Devendra Garwa

I’m an IIT Roorkee graduate building end-to-end analytical solutions using machine learning, predictive modeling, and data analytics. I enjoy turning messy data into decision-ready systems across domains like logistics, credit risk, customer analytics, inventory optimization, and geophysical processing.

In my Machine Learning Engineer Intern role at Technocolabs Softwares Inc., I developed a predictive delivery time estimation model using Python, Pandas, and NumPy on 35K+ logistics records. I engineered 20+ domain-specific features and evaluated models including Lasso, Random Forest, XGBoost, and CatBoost, finalizing with CatBoost.

My project work reflects how I approach problems: strong data preparation, interpretable modeling, and practical outcomes. For example, I built a credit risk classification and risk scoring model on 25K+ records, and I used SMOTE and decision-threshold tuning for high-risk flagging.

I also bring explainability and performance-focused tooling to my work—using SHAP for insight and Scikit-learn for robust evaluation. Beyond pure ML, I’ve built SQL + Power BI analytics for inventory KPIs and constructed a Python pipeline for seismic attenuation analysis, including anomaly detection and interactive 3D mapping.