Sanchari Biswas is a PhD student in Computer and Information Sciences at Temple University, where she is advised by Prof. Chiu C. Tan. Her research focuses on understanding and improving the reliability of Large Language Models (LLMs). She studies how these systems generate information, why they sometimes produce hallucinations or inconsistencies, and how structured reasoning and evaluation frameworks can help make AI systems more trustworthy. Her work lies at the intersection of trustworthy AI, language model reasoning, and security-oriented evaluation of modern AI systems.
Research Interests
- Hallucination Detection and Mitigation in Large Language Models
- Trustworthy, Safe, and Reliable AI Systems
- Multilingual Robustness and Cross-Lingual Security
- Evaluation Frameworks Inspired by Formal Verification
- Policy-aware and Guardrail-aware AI Reasoning
Current Focus
Sanchari’s current research investigates systematic approaches to identifying and analyzing hallucinations produced by large language models across domains such as security, education, and healthcare. Her work explores structured intermediate reasoning representations, constraint-based verification techniques, and scalable benchmarking methods designed to evaluate whether model outputs remain consistent, reliable, and aligned with user intent and safety policies.
About Me
Outside of research, Sanchari enjoys exploring creativity in different forms. She loves reading fiction, experimenting with cuisines from around the world, and discovering cultural connections through food. Cooking serves as both a creative outlet and a way to unwind after research. She is also passionate about music and enjoys singing Rabindrasangeet, finding it to be a grounding and reflective practice alongside the fast-paced world of academic research.