Ameesh Mithanthaya

Smart News Synthesizer

• Built a RAG-based news verification and summarization system integrating credibility classification and semantic retrieval. • Engineered the NLP pipeline using BERT, FAISS, and Sentence Transformers with latency optimized to 3–4 seconds. • Achieved a BERT classifier F1-score of 0.71 and reduced article reading time by 60%; system published as an Indian Patent.

Fora AI: Autonomous Research & Intelligence Engine

• Architected an autonomous research engine in LangGraph utilizing custom Model Context Protocol (MCP) tool servers for web scraping and text extraction, cutting manual research turnaround time by 75%. • Implemented cyclical reflection state graphs with automated critique gates, enabling self-correction over failed search queries and driving source attribution accuracy to 98%. • Built an asynchronous FastAPI backend with SQLite state checkpointing, supporting background job polling and delivering structured briefings with live citations under 45 seconds.

Pharma Watch: Drug Adverse Event Trend Analyzer

arma Watch: Drug Adverse Event Trend Analyzer | Python, Streamlit, SQLite, Pandas, Plotly • Developed a pharmacovigilance analytics dashboard to analyze 100,000+ FDA adverse event reports across thousands of drugs and adverse events. • Built ETL pipelines for FAERS datasets, performing data cleaning, validation, standardization, and transformation. • Designed an interactive Streamlit dashboard with 15+ visualizations to identify reporting trends, demographic patterns, and drug-event relationships.