GenAI/NLP

IntelliQA: RAG-Based Document QA Assistant

SupportIQ: Phase I&II (LLM Fine-Tuning)

SupportIQ: Phase III (Rasa Chatbot)

ML

A NOVEL ELASTIC PERIODIC ACTIVATION FOR DEEP NEURAL NETWORKS

Dynamic Price Optimization Using Elasticity of Demand

Insurance Pricing Forecast Using XGBoost Regressor

Developed an XGBoost Regressor model to predict healthcare charges, optimizing insurance premium strategies based on customer features like age, BMI, smoking status, and region. Conducted exploratory data analysis (EDA) and correlation tests to inform model development, improving prediction accuracy by replacing a baseline linear regression model with XGBoost. Enhanced model performance using Sklearn's Pipeline and BayesSearchCV for hyperparameter optimization, presenting results in an accessible format for non-technical stakeholders.

Opioid Risk Factor Identification using Machine Learning

Developed a machine learning framework using PROMIS scale data from UT Southwestern Medical Center to identify risk factors for opioid-based disorders. Applied Logistic Regression to predict high opioid risk, achieving 90% recall and 40% precision. Interpreted model outputs to identify key risk factors and optimized prediction thresholds, improving the detection of high-risk individuals.

Statistics

Experimental Design for RFID Tag Performance Optimization