Kathakoli Sengupta

Kathakoli Sengupta

Stony Brook University

I am a first-year PhD student in Synthetic Perception and Learning Lab (SPELLAB) under the Computer Vision Lab advised by Prof. Paola Cascante-Bonilla. Before starting my PhD, I worked as a Computer Vision Scientist at Wicket LLC. During my masters, I had the priviledge of working with Prof. Eshed Ohn Bar.

Currently, My research aim is to explore compositional and spatial reasoning of VLMs in 3D spaces. I am also integrating embodied AI into 3D generated environments to explore the reasoning abilities of such agents.

🔥 News

📝 Publications

Preprint
Preprint

SceneCritic: A Symbolic Evaluator for 3D Indoor Scene Synthesis
Preprint

Kathakoli Sengupta, Kai Ao, Paola Cascante-Bonilla

Project | Code

NeurIPS 2024
NeurIPS2024

Text to Blind Motion
NeurIPS 2024 (Poster)

Hee Jae Kim, Kathakoli Sengupta, Masaki Kuribayashi, Hernisa Kacorri, Eshed Ohn-Bar

Project | Code

ECCV 2024
ECCV2024

UniLCD: Unified Local-Cloud Decision-Making via Reinforcement Learning
ECCV 2024 (Poster)

Kathakoli Sengupta, Zhongkai Shangguan, Sandesh Bharadwaj, Sanjay Arora, Eshed Ohn-Bar, Renato Mancuso

Project | Code

💻 Experience

  • 2026.05 - Now, Research Assistant, Stony Brook University (Advisor: Prof. Paola Cascante-Bonilla).
  • 2025.08 - 2026.05, Teaching Assistant, Stony Brook University (CSE590 Vision and Language Models).
  • 2024.07 - 2025.05, Computer Vision Scientist, Wicket LLC.
  • 2023.03 - 2024.05, Research Assistant, Boston University College of Engineering, H2X Lab (Advisor: Prof. Eshed Ohn-Bar).
  • 2023.09 - 2023.12, Teaching Assistant, Boston University (EC-518 Robot Learning).
  • 2022.10 - 2023.05, Research Assistant, Boston University School of Medicine, Bio-Imaging & Informatics Lab (Advisor: Prof. Bang-Bon Koo).
  • 2021.09 - 2021.11, Machine Learning Intern, University of Calcutta, Rajabazar Campus (Advisor: Prof. Rajarshi Gupta).
  • 2021.08 - 2021.10, Data Science and Machine Learning Intern, Indian Institute of Technology, Delhi (Advisor: Prof. Abhijit Majumdar).

📖 Educations

  • 2025.08 - Now, PhD in Computer Science (GPA 4.00), Stony Brook University, New York, USA.
  • 2022.09 - 2024.05, Masters in Artificial Intelligence (GPA 3.96), Boston University, Boston, USA.
  • 2018.07 - 2022.05, BTech in Electronics and Communication Engineering (GPA 3.95), VIT, Vellore, India.

🎖 Honors and Awards

  • 2025.08 Dept. of Computer Science John Hennessey Scholarship, Stony Brook University, New York, USA.
  • 2019.01 Merit Scholarship, School of Electronics Engineering, VIT, Vellore, India

Selected Projects

  • A Vision-Language Approach to Efficient Scene Layout Generation
  • Learning Spatial Representation for Efficient Robot Navigation(Masters Thesis)
  • Person Following LIMO Robot
  • NeRF Editing with Geometric Processing
  • Early‑Exit Inspired Dynamic Neural Network For OOD Satellite Imaging
  • Driver Accident Prevention System (Bachelors Thesis)

Co-Curricular

  • 2021.01 - 2021.12, Chairperson, IEEE SPS VIT.
  • 2021.06 - 2021.07, Organiser, HackX: Unveil Your X-factor, IEEE SPS VIT.
  • 2020.09 - 2020.10, Event Coordinator, Building Chatbot: Expertise from Scratch, graVITas VIT.
  • 2020.03, Coordinator, Reboot, IEEE Robotics and Automation Society, VIT Vellore.
  • 2020.01 - 2020.12, Technical Head, IEEE Robotics and Automation Society, VIT Vellore.
  • 2019.07 - 2020.07, Program Representative, School of Electronics Engineering, Vellore Institute of Technology.
  • 2018.12 - 2020.12, Core Committee Member, IEEE SPS VIT.
  • 2018.12 - 2019.12, Core Committee Member, IEEE Robotics and Automation Society, VIT Vellore.