About Me

I am a second year CS PhD student at Khoury College of Computer Science at Northeastern University in Boston. I am advised by Prof Mai ElSherief.

Before starting my PhD, I was a Software Engineer with the Selection Monitoring and Catalog Systems organization at Amazon in Seattle. Prior to that I was an MS CS student at Georgia Tech. At Georgia Tech, I was advised by Prof Srijan Kumar in the CLAWS Lab. I also dabble as a Research mentor with SimPPL.

Research Interest

I am interested in designing AI that are safer, more controllable, and socially aligned. My research areas broadly focus on the following:

🤹🏼 Personalization

• How do models internalize and represent human personas? I explore how latent subspaces can be used to enable more precise behavioral control.
• How does user context influence model behavior? My research examines how context drives variation in empathy and how models respond to implicit cues such as dialect or identity.
• How do LLMs build internal “maps” of the world? I investigate how geographic and spatial knowledge is structured within a model’s weights and how this impacts downstream reasoning.

🛡️ AI Safety and Social Alignment

• Can we steer models toward safer behavior without retraining? I use activation patching and causal mediation analysis to identify internal circuits and develop inference-time methods for bias mitigation.
• How can we make model reasoning more robust? I research ways to steer chain-of-thought processes using light weight interventions to ensure that a model’s logical “thought” traces are consistent and aligned with human expectations.
• How do we identify adversarial and negative LLM behavior? I study what triggers and how to stop model’s harmful social behavior.

🤝🌐 Computational Social Science

• How can AI support healthier online communities? My work focuses on building tools that identify nuanced social phenomena, such as detecting dog whistles and analyzing the reasoning behind hate speech.
• How does AI interpret intent in digital spaces? I explore how NLP frameworks and prompt optimization can be applied to study rampant motherhood burnout as expressed in reddit communities

Academic Service and Groups

Groups

• CSG (Computation for Social Good) Lab @ NEU
• Geodesic Research
• MAIA
• SimPPL
• CLAWS

Teaching

• TA for CS 5200: Database Management Systems

• TA for CS 4100: Foundations of AI (Spring 2025) with Prof Chris Amato

• TA for CS 3600: Intro to AI, with Prof James Rehg (Spring 2022) and Prof Mark Reidl (Fall 2022)

Talks

• Slides of my lecture on Advanced Topics in AI

• Presentation and Slides at SafeGenAI workship at NeurIPS on Who Speaks Matters: Analysing the Influence of the Speaker’s Ethnicity on Hate Classification