Sreenivas REDDY

• Examined more than 500,000 customer transactions and 10,000 credit profiles through Python, SQL, Pandas, and NumPy to determine spending patterns and demographic insights.
• Developed customer segmentation and credit risk models by evaluating credit scores, utilization ratios, and spending behaviours for age group 18–25, identifying high-potential segment for targeted marketing.
• Constructed robust data preprocessing pipelines to handle missing values, duplicates, and outliers for producing high-quality datasets in analytics.
• Executed A/B testing experiments with power analysis and optimal sample size determination, providing data-driven insights that guided marketing strategies.

Credit Risk Modeling & Credit Score Prediction

• Created an end-to-end credit risk modelling application using Python, Scikit-learn, Pandas, NumPy, and Streamlit to predict loan default probability and standardize credit scores, enhancing risk assessment processes.
• Implemented a complete machine learning pipeline, including feature engineering and model deployment with Joblib for scalable financial risk assessment.

Vehicle Damage Detection using Deep Learning (PyTorch, ResNet50 & Streamlit)

• Developed an AI-powered Vehicle Damage Detection application using PyTorch, ResNet50 Transfer Learning, TorchVision, and Streamlit to classify vehicle damage into 6 categories,(Front/Rear Normal, Breakage, and Crushed). implementing an end-to-end image preprocessing, feature extraction, and real-time inference pipeline for uploaded vehicle images.
• Fine-tuned a pre-trained ResNet50 model by retraining the final classification layers while freezing backbone layers, achieving ~80% validation accuracy on a dataset of ~1,700 vehicle images, and deployed the solution as an interactive Streamlit web application with drag-and-drop image upload and instant predictions.