Avishree Khare

prof_pic.jpg

Hi! I am a third-year PhD student in Computer Science at the University of Pennsylvania, where I am fortunate to be advised by Prof. Rajeev Alur, and sponsored by a fellowship from Toyota.

I am broadly interested in research topics at the intersection of Machine Learning and Formal Methods. My current research uses ideas from Formal Logic to improve the reliability and robustness of AI systems. Some problems that I am working on include:

  • How can we enhance video understanding through symbolic reasoning?
  • How can we ensure that generative models follow rules strictly?
  • How can we leverage generative models for complex reasoning (e.g., in software engineering and potentially, mathematics)?

Prior to joining Penn, I was a Research Fellow with the PROSE group at Microsoft . I also spent a year as a Software Engineer at Google. Before that, I graduated with a B.E in Computer Science from BITS Pilani Goa, India.

In my spare time, I enjoy running, (pen) sketching and baking bread!

I am looking for internships in Summer 2026. My CV is here, please reach out!

publications

  1. Instruction Following by Boosting Attention of Large Language Models
    Vitoria Guardieiro*, Adam Stein*, Avishree Khare*, and Eric Wong
    (Spotlight) Mechanistic Interpretability Workshop, NeurIPS , 2025
  2. LogSTOP: Temporal Scores over Prediction Sequences for Matching and Retrieval
    Avishree Khare, Hideki Okamoto, Bardh Hoxha, Georgios Fainekos, and Rajeev Alur
    A preliminary version of this work will be presented at the NextVid Workshop, NeurIPS, 2025
  3. Logicbreaks: A Framework for Understanding Subversion of Rule-based Inference
    Anton Xue*, Avishree Khare*, Rajeev Alur, Surbhi Goel, and Eric Wong
    In The Thirteenth International Conference on Learning Representations (ICLR), 2025
  4. Boss LLM: Adaptation via No-Regret Learning
    Yu Feng*, Avishree Khare*, Nghia Nguyen*, and Sikata Bela Sengupta*
    In Workshop on Scaling Self-Improving Foundation Models without Human Supervision, ICLR, 2025
  5. Understanding the Effectiveness of Large Language Models in Detecting Security Vulnerabilities
    Avishree Khare*, Saikat Dutta*, Ziyang Li, Alaia Solko-Breslin, Mayur Naik, and Rajeev Alur
    In Proceedings of the 2025 IEEE Conference on Software Testing, Validation and Verification (ICST), 2025
  6. Grace: Language models meet code edits
    Priyanshu Gupta*, Avishree Khare*, Yasharth Bajpai, Saikat Chakraborty, Sumit Gulwani, Aditya Kanade, Arjun Radhakrishna, Gustavo Soares, and Ashish Tiwari
    In Proceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (FSE), 2023

*equal contribution