Shanka Subhra Mondal

Hi! I am a Senior Research Scientist at the Accenture Center for Advanced AI in Mountain View, California, where I joined in 2026. My research focuses on multimodal agents, multi-agent reinforcement learning, and visually grounded reasoning for long horizon tasks.

I completed my Ph.D. in Electrical and Computer Engineering at Princeton University in 2025, advised by Jonathan D. Cohen (Ph.D. thesis). My doctoral research drew on cognitive science and neuroscience to develop neural networks with human-like systematic and sample-efficient generalization, with a focus on object-centric representations and abstract visual reasoning.

During the summers of 2023 and 2024, I interned at Microsoft Research, New York, with Ida Momennejad, developing a modular, multi-agent architecture using large language models to improve multi-step planning. This work was published in Nature Communications in 2025. At Princeton, I also worked on projects with Sebastian Seung and Tom Griffiths.

Before Princeton, I earned my B.Tech. at the Indian Institute of Technology, Kharagpur, where I worked with Debdoot Sheet at the intersection of deep learning and biomedical imaging. I also interned at Adobe Research, Bangalore, with Subrata Mitra, developing a deep reinforcement learning scheduler to improve resource utilization and application performance.