Machine learning framework for predicting placebo responsiveness using a 19-feature psychological and clinical dataset (~13,750 samples). Applied SMOTE, stratified 5-fold cross-validation, and compared Logistic Regression, Random Forest, and SVM models, achieving 82% accuracy and a ROC-AUC of 0.91 with Random Forest. Presented at IEEE CONIT 2026 (Accepted for Publication).

Full-stack e-commerce platform developed for a business client featuring a product catalogue, Razorpay payment integration, Google OAuth authentication, admin order management, and automated contact inquiry system. Responsible for end-to-end development from client requirements to deployment.

Full-stack university support platform for managing campus issues, complaints, and student feedback through a centralized dashboard. Implemented secure university email validation, issue lifecycle tracking, responsive Tailwind UI, dark mode, LocalStorage-based data management, and a rule-based chatbot assistant to enhance navigation and user experience.

Production-style mobile banking backend built with Flask featuring a complete DevOps pipeline. Automated CI/CD using Jenkins and GitHub Actions, containerized with Docker, deployed on Kubernetes, and provisioned AWS infrastructure using Terraform. Includes automated testing, Docker image builds, and reproducible cloud deployments.

Complete infrastructure automation project implementing Infrastructure as Code with Terraform, Kubernetes container orchestration, GitHub Actions CI/CD pipelines, and Ansible configuration management. Supports automated multi-environment deployments following DevOps best practices.

Natural language processing application for detecting clickbait news headlines using TF-IDF feature extraction and multiple machine learning classifiers including Logistic Regression, Naive Bayes, and SVM. Includes an interactive Streamlit web interface for real-time predictions.