About Me

I am a PhD student in Computer Science at the University of Maryland, Baltimore County, advised by Dr. Tejas Gokhale. I also closely collaborate with Dr. Kowshik Thopalli and Dr. Vivek Narayanaswamy from Lawrence Livermore National Laboratory. My research interests span across continual learning and concept-based interpretability with key focus on harnessing concept-based respresentations for adapting to novelty and variations.

Before starting my PhD, I was a Research Associate at IIT Hyderabad under the guidance of Dr. Vineeth Balasubramanian, where I worked on Generalized Class Discovery (GCD) and Open World Object Detection(OWOD) problems.

I completed my Bachelor of Engineering in Electronics and Communication Engineering from KLE Technological University, where I was advised by Dr. Uma Mudenagudi

I am open for collaborations—reach out with your cool ideas to catch up and discuss. Explore my publications or contact me to start a conversation.

News

  • June 2026: Our paper "On the Role of Mechanistic Interpretability for Vision-Language Prompt Learning" has been accepted to the ICML Workshop on Mechanistic Interpretability.
  • May 2026: Gave a talk at COEIT Research Day on "Sequentially Acquiring Concept Knowledge to Guide Continual Learning".
  • March 2026: Serving on the Program Committee for CoLLAs 2026 and NeurIPS 2026.
  • November 2025: Serving on the Program Committee for CVPR 2026.
  • Summer 2025: Research Intern, Lawrence Livermore National Laboratory (Machine Intelligence Group).
  • June 2025: Program Committee — AAAI, CoLLAs; ELAMI workshop @ MICCAI.
  • May 2025: SACK: Sequentially Acquiring Concepts to Guide Continual Learning — accepted to Workshop on Visual Concepts @ CVPR.
  • August 2024: Joined CVG @ UMBC as a PhD student.
  • February 2024: Our paper "Novel class discovery for representation of real-world heritage data as neural radiance fields (student abstract)" accepted at AAAI 2024.
  • January 2024: Our paper "A benchmark grocery dataset of real-world point clouds from single view" accepted to 3DV 2025.