Projects
Supply Chain Anomaly Detection for Delivery Fraud
Developed ML models (Random Forest, SVM, Transformer architectures) to detect logistics fraud, identifying 7% anomalies with 91.8% accuracy.
Multi-head Regression for Order Fulfillment Prediction
Trained a multi-head regression model in PyTorch to predict fulfillment time, order profit, and delay likelihood, optimizing MSE and BCE losses, achieving high predictive accuracy across all targets.
AI-Powered Mental Health Chatbot with Context-Aware Conversations
Built and fine-tuned a multi-turn chatbot using Hugging Face Transformers, optimizing mental health support responses based on BLEU evaluation.
Insurance Claim Fraud Detection using Machine Learning
Built ML pipeline on a real-world insurance dataset (with 1000+ records and 40+ features) using Random Forest, XGBoost, SMOTE balancing, and feature scaling, achieving accurate fraud detection and identifying key behavioral risk factors.
Caregiver Stress Classification using Machine Learning
Designed and conducted a socio-demographic survey, then trained SVM models on the collected data to predict caregiver stress levels, achieving 98% prediction accuracy.
Published Articles/PoV at ZS
Harnessing GenAI to transform IC Design | Emerging trends in IC for Healthcare Ecosystem models | Designing IC Plans for Omni-channel Marketing Teams | Handling IC during Natural Disasters