Mahsa Ghanbarpour Mamaghani

Assistant Director-data Scientis @Liberty Mutual Insurance

Boston, MA, US

Early Injury Detection Model:Integrating an LLM to predict injury scores from FNOL call transcripts, and then incorporating those predictions alongside other variables in a ML model.USC Summarization Pipeline:• Developed LLM-based NLP pipeline for automated summarization of unstructured claims data.WC Early severity model :Built end-to-end supervised learning to predict claims with high severity in early ages of claims including building pipeline to process data, fit NLP and Machine Learning modelsLarge Loss Prediction Model:• Built end-to-end supervised learning pipeline (NLP, CNN, XGBoost) to predict large loss claims (>$1M).Auto Ultimate and GL Ultimate Modeling:• Led feature engineering and CNN model development for ultimate loss projections and IBNR automation.

Northeastern University

Doctor of Philosophy (Ph.D.), Industrial Engineering

University of Tehran

Master's Degree, Industrial Engineering

Sharif University of Technology

Bachelor's Degree, Industrial Engineering