About Me
Hey, I am Ishan Kavathekar! I'm a Research Associate at Adobe Research India. I am a dual-degree Computer Science graduate from International Institute of Information Technology, Hyderabad where I was an undergraduate researcher at Precog Lab, working under the guidance of Dr. Ponnurangam Kumaraguru.
Previously, I interned at Microsoft Research, mentored by Tanuja Ganu, and at Adobe Research, where I worked under Balaji Vasan Srinivasan and Aishwarya Agarwal.
My primary research interests lie in AI safety and in developing robust, trustworthy AI systems. I am currently working on assessing the capabilites in LLMs along with robustness of LLM-based agents and multi-agent systems. I am particularly interested in exploring how these systems behave in adversarial settings and how we can make them more reliable and trustworthy.
I am actively looking for potential research collaborations in the areas of Agentic Safety, Emergent Behaviors in AI models and AI Saefty in general. If your interests align with mine or if you’d simply like to discuss research ideas, please feel free to reach out!
Education
News
Joined Adobe Research India as a Research Associate!
Defended my MS Thesis titled "Towards Trustworthy LLM Systems: Safety, Reliability, and Robustness"
TAMAS accepted at ACL 2026 Main Conference! 🎉
"Exposing Weak Links in Multi-Agent Systems under Adversarial Prompting" accepted at AAMAS SE Workshop 2026! 🎉
Selected as one of 40 participants globally for the International AI Evaluation Program (≈6% acceptance rate)! 🎉
TAMAS: A Dataset for Investigating Security Risks in Multi-Agent LLM Systems" was accepted at ICML-MAS workshop 2025! 🎉
Started my research internship at Adobe Research, Bengaluru in the Multi-Modal Content Group.
"Small Models, Big Tasks: An Exploratory Empirical Study on Small Language Models for Function Calling" was accepted at EASE 2025 AI Models and Data Evaluation track! 🎉
Presented "Counter Turing Test" at SUMEval-2 Workshop @ Coling 2025.
Started my research internship at Microsoft Research, Bengaluru under Tanuja Ganu.
"Counter Turing Test (CT^2): Investigating AI-Generated Text Detection for Hindi -- Ranking LLMs based on Hindi AI Detectability Index (ADI_{hi})" was accepted at EMNLP 2024 Findings! 🎉
Selected Publications
TAMAS: Benchmarking Adversarial Risks in Multi-Agent LLM Systems
Ishan Kavathekar*, Hemang Jain*, Ameya Rathod, Ponnurangam Kumaraguru, Tanuja Ganu
ACL 2026 Main and ICML-MAS Workshop 2025
Counter Turing Test (CT^2): Investigating AI-Generated Text Detection for Hindi - Ranking LLMs based on Hindi AI Detectability Index (ADI_hi)
Ishan Kavathekar, Anku Rani, Ashmit Chamoli, Ponnurangam Kumaraguru, Amit P. Sheth, Amitava Das
EMNLP Findings 2024