About Me
Hi! I’m Rasika. I’m a 2nd year PhD student at Indiana University, Bloomington, where I am fortunate to be advised by Dr. Jisun An and Dr. Haewoon Kwak. I’ve also had the priviledge of working closely with Dr. Yong-Yeol Ahn.
My research on social simulation of multi-LLM agent systems with a particular interest in social norms and beliefs. I explore ideas through mechanism simulations, empirical works with social media platforms, and mechanistic interpretability techniques. Broadly, I’m passionate about human-AI alignment and interaction. I explore how LLMs can be used as simulation devices to study interactions and uncover relationships between the individual and the collective.
Currently, I’m expanding my work to examine how LLMs can support content moderation and how understanding the internal workings of these models can lead to better value-aligned systems—aligned in beliefs, behaviors, and social expectations.
Prior to starting my Ph.D., I completed a Master’s in Data Science at Indiana University. Before that, I earned a Bachelor of Technology in Electrical and Electronics Engineering from Manipal Institute of Technology, India.
Outside of research, I enjoy reading (mostly fantasy), arranging a cappella music, and dabbling in new activities whenever I get the chance.
News
- July 06, 2026:
Presented our work “PluRule” as a poster with Zoher Kachwala at ACL 2026 in San Diego, CA, USA.
- June 22, 2026:
I attended the Lake Como Summer School on Computational Social Science for the week and gave a student presentation on my work, “Norm Emergence in LLM Societies.”
- April 15, 2026:
I’ll be presenting my latest work ‘SoNoLiSi: Simulating the Social Norm Lifecycle with Generative Agents’ at Midwest Speech and Language Days 2026 at Univerity of Illinois, Urbana-Champaign (UIUC), 2026.
- April 09, 2026:
Our paper ‘Federating Governance: Scaling Communities Rules on Mastodon’ has been accepted to CSCW 2026. Shout out to collaborators Dr. YY Ahn and Dr. Bao Tran Truong. See you in Salt Lake City, Utah. Arvix link:
- April 08, 2026:
I passed my qualifying exam! I am now a PhD candidate!
- April 07, 2026:
Our paper “PluRule: A Challenging Benchmark for Detecting Rule Violations of Pluralistic Communities on Social Media” has been accepted to ACL 2026. Shout out to collaborators Dr. Zoher Kachwala and Dr. Bao Tran Truong. See you in San Diego!
- March 16, 2026:
I got accepted to the Lake Como Summer School on Computational Social Science! Excited to visit the beautiful city of Como and collaborate with fellow CSS folks!
- January 16, 2026:
Our preprint “XChoice: Explainable Evaluation of AI-Human Alignment in LLM-based Constrained Choice Decision Making” is available on Arvix. (Under Review at ACL ARR)
- July 21, 2025:
I will be at [\(IC^{2}S^{2}\)] in Norrkoping, Sweden from 21st July to 24th July. Come catch my talk on “ Scaling Rules Across Mastodon Instances” at 11 am (Social Media & Networks track), 24th July 2025.
- April 18, 2025:
I served as a reviewer for the first time! Exciting!
- April 18, 2025:
My work with Dr. Bao Tran Truong, titled “Scaling of Community Norms on Mastodon” was accepted at at [\(IC^{2}S^{2}\)].
- April 15, 2025:
I presented my work “From Team Science to AI Teams: Understanding the Social Dynamics of Multi-Agent Reasoning” at University of Notre Dame for MSLD 2025!
- March 25, 2025:
My abstract on Agentic Teams (the application of Team Science on Agentic Teams) got accepted at Midwest Speech and Language Days (MSLD) 2025! See you in Notre Dame!
- August 26, 2024:
I started my PhD in Informatics at Indiana University Bloomington in the Complex Networks and Systems track. I will be working with Prof. Yong-Yeol Ahn (yylab) and with Prof. Jisun An and Prof. Haewoon Kwak (SODA lab).
- August 23, 2024:
Our work titled “Digital epidemiology of Prescription Drug References on X: Neural Network Topic Modeling and Sentiment Analysis” was published in JMIR.
Ongoing Projects
- Multi-LLM Agent Systems for Social Simualtions
– Social Norms
– Belief Resonance
- Mechanistic Intepretability and Internal Representations of beliefs in LLMs
Using mechanistic interpretability methods to see under the hood of LLMs to investigate if it embodies persona, values and beliefs they we think it does, both individually and in a group.
- Platform Self-Governance
– Characterizing Bluesky Self-Governance System
Examining the characteristics of Bluesky’s self-governance system to understand the nuances of the new growing social media platform that many are migrating to.
- AI Safety
– Coordination Failures
– Normative Behavior Failures