I recently completed my Ph.D. in Computer Science from the University of Southern California (USC), where I was advised by Prof. Morteza Dehghani. My research focuses on Large Language Models (LLMs), Transfer Learning, and Dialogue Systems. I am particularly interested in understanding, predicting, and evaluating the behavior of LLMs to enable their safe, reliable, and effective use.
My work spans both the development of methods for improving the machine learning pipeline—such as model selection, synthetic data generation, user simulation, and evaluation—and the application of LLMs to specialized domains. I study how language models can be adapted to specialized domains while remaining trustworthy and accessible to both experts and non-experts, including:

• Developing LLM-based methods for law enforcement and public safety (with Morteza and the CALM Lab)
• Detecting software vulnerabilities using LLMs (with Sima)
• Building AI systems for math (topology) that extract and formalize topological knowledge from mathematical literature (with Gunnar and Rahul)
• Developing user simulation methods for personalized dialogue systems (with Bahar and Sheshera)
  

I’m always happy to discuss my research, explore new ideas, and connect with researchers. If our research interests overlap, feel free to reach out!

Work Experience

• Research Intern @ Microsoft, Summer 2025
  
  • Supervisor: Bahareh Sarrafzadeh
• Graduate Student Researcher @ Google Research, Fall 2022 - Spring 2023
  
  • Supervisor: Hakim Sidahmed

Selected Publications

• A Systematic Analysis of Base Model Choice for Reward Modeling
  K Ahrabian, P Jandaghi, N Mokhberian, P Karimireddy, J Pujara
  In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing

• Faithful Persona-based Conversational Dataset Generation with Large Language Models
  P Jandaghi, X Sheng, X Bai, J Pujara, H Sidahmed
  In Findings of the Association for Computational Linguistics: ACL 2024

• Reflect, Not Reflex: Inference-Based Common Ground Improves Dialogue Response Quality
  P Zhou, H Cho, P Jandaghi, DH Lee, B Lin, J Pujara, X Ren
  In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing

• FETA: A Benchmark for Few-Sample Task Transfer in Open-Domain Dialogue
  A Albalak, YL Tuan, P Jandaghi, C Pryor, L Yoffe, D Ramachandran, L Getoor, J Pujara, W Wang
  In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing