Fine-tuned roberta-base classifier on the LIAR dataset. Aaccepts multiple input types text, URLs, and PDFs and outputs a prediction with a confidence score. It also leverages google/flan-t5-base to generate explanations and uses an Agentic AI with LangGraph to orchestrate agents for planning, retrieval, execution, fallback, and reasoning.

AI research for bias mitigation and truth verification - building fair, transparent systems that combat misinformation while ensuring equitable outcomes across all demographics.