ML/NLP Research Intern at MInterns (Dec 2025–Mar 2026)
TextBuilt and evaluated an NLP-based political security threat prediction framework using a hybrid lexicon approach combined with ensemble machine learning. Created an NLP preprocessing pipeline on Twitter text datasets, including tokenization, stopword removal, lemmatization, and TF-IDF vectorization. Implemented and benchmarked multiple text classifiers and measured performance, including a stacking classifier achieving 99.4% accuracy. • Developed and benchmarked Naive Bayes, SVM, Decision Tree, VotingClassifier, and StackingClassifier models. • Performed feature extraction using TF-IDF and topic modeling via LDA. • Deployed predictions in a Flask-based multilingual web application for real-time text inference. • Produced end-to-end system outputs with automated multilingual translations for usability.