Oral
ECCV 2024 Best Paper Honorable Mention
Circumventing Concept Inhibition in Text-to-Image Diffusion Models
About
I am a Senior Researcher on the Vision Intelligence Team at Samsung Research America, where I work on Multi-Modal Language Models, Agentic Memory and Reasoning, and Efficient Multi-Modal Systems.
I received my Ph.D. in Computer Science from Boston University in 2024, advised by Prof. Kate Saenko. My doctoral research focused on Explainable AI — developing feature attribution methods for Computer Vision and investigating concept manipulation in Text-to-Image Diffusion Models. I also spent two summers as a Machine Learning Research Intern at Adobe Research.
Experience
Senior Researcher, AI Research
Samsung Research America · Mountain View, CA
Working on Efficient Vision-Language Systems.
Machine Learning Research Intern
Adobe Research · College Park, MD
Developed novel explainability methods for object detection (D-RISE) and image similarity models. Received the Adobe Research Fellowship in 2020.
Research & Teaching Assistant
Boston University · Boston, MA
Conducted research in Explainable AI and computer vision under Prof. Kate Saenko. Published at top-tier venues including CVPR, ECCV, and NeurIPS workshops.
Education
Ph.D. in Computer Science
Boston University · Boston, MA
Advised by Prof. Kate Saenko. Research in Explainable AI, concept manipulation in diffusion models, and feature attribution methods for computer vision.
M.S. / B.S. in Computer Science & Applied Mathematics
Belarusian State University · Minsk, Belarus
Master's Program in Computer Science
Yandex School of Data Analysis · Minsk, Belarus
Publications Google Scholar ↗
Oral
Oral
Oral
Oral
Side Projects
Fun things I've built outside of research.
💡
Project Name
Short description of the project, what it does, and why it's interesting.
Python
PyTorch
🚀
Project Name
Short description of the project, what it does, and why it's interesting.
JavaScript
React
🎨
Project Name
Short description of the project, what it does, and why it's interesting.
Python
Diffusion