Khushboo Singh

Performed longitudinal LDA analysis on 10,108 EMA texts to identify affect–alcohol and stress–alcohol trajectories and visualized 30 topic clusters per group to inform behavioral patterns.

Built and benchmarked a Transformer-based pipeline on DS4UD to predict 43 psychological outcomes, achieving r≈0.6 for mental health and r>0.7 for personality traits via stratified 10-fold evaluation.

Analyzed speech data with LLMs to identify linguistic markers of PTSD, depression (PHQ-9), and cognitive function (MoCA) under the World Trade Center Health Program to support clinical screening.

Engineered a multi-LLM annotation framework to detect PII and policy contradictions in medical app privacy policies, achieving 85% accuracy in GDPR violation identification.

Developed and deployed 5+ ML-powered mobile apps serving 10K+ users, integrated recommendation and personalization systems that increased engagement up to 30%, and built a LangChain chatbot automating 70% of support queries.

Developed a cyber-physical system for smart manufacturing with XGBoost predictive maintenance models and a digital twin simulator, reducing manual intervention by 90% and enabling real-time analytics.

Grade: 3.85/4

Activities and societies: Vice President, Sky Campus Happiness Club; Secretary, Indian Graduate Student Association; Software Developer, Stony Brook Robotics Team

M.S. in Computer Science with thesis work evaluating transformer LM representations for author states and traits; coursework included NLP, Machine Learning, AI, and Algorithms.

Grade: 9.36/10

B. Tech. in Electrical Engineering and Computer Science with coursework in operating systems, algorithms, computer architecture, networks, and neural networks.