NoKe is a machine learning-powered fake news detector featuring a full MLOps pipeline and a Chrome extension. It leverages both traditional models like Random Forest (via Scikit-learn) and advanced transformer models like RoBERTa (using HuggingFace and PyTorch Lightning). The system includes a modular NLP pipeline with custom featurizers, TF-IDF vectorization, and transformer-based embeddings. Evaluation is done using metrics like F1 score, AUC, and SHAP for interpretability. DVC ensures data versioning, MLflow tracks experiments, and Great Expectations handles data validation. The model is served via a FastAPI backend, containerized with Docker, and integrated into CI workflows using GitHub Actions.