I developed a Dual-Engine Fake News Detection & Real-Time Live Verification System that combines machine-learning classification with live news verification to return REAL, FAKE, or UNVERIFIED outcomes.
I built its NLP classification pipeline with TF-IDF and Multinomial Naive Bayes, Linear SVM, and Logistic Regression, optimized through 5-fold Stratified GridSearchCV. On 3,344 unseen stratified test samples, it achieved up to 93.39% accuracy.
The system also checks Google News RSS using query extraction and TF-IDF cosine similarity, drawing on 60+ verified sources. Its decision layer lets verified live matches override conflicting model predictions.
As a Full Stack Developer Intern at Future Full Stack Developer, I supported the company’s local web platform for its social media application. I developed and modified frontend features using HTML and worked on backend functionality using Java and Python.
