Niladri Shekhar Dutt
I'm a second-year ELLIS PhD candidate (start: Oct'24) at University College London (UCL), sponsored by Adobe Research. I'm advised by Niloy Mitra (UCL) and Duygu Ceylan (Adobe). I'm currently interning again as a Research Scientist at Adobe Research with Jakub Fiser on video generation.
My research studies visual representation and multimodal learning — how the visual world should be represented in tokens, if generative vision models represent structured plans, and whether explicit structured intermediate representations improve compositional generalization in foundation models.
Previously Research Scientist Intern at Adobe (with Xuelin Chen) · Research Intern at NVIDIA · Visiting Student at UC Berkeley · Research Engineer at Ready Player Me (acquired by Netflix) · Founding Engineer at NFTPort (raised $26M, now Galadriel)
News
scroll ↓- Nov 2026 Co-organizing the BMVA Symposium on World Models. upcoming
- Jul 2026 Invited talk at Voxel51 & Imperial College London.
- Feb 2026 Invited talk at the Stanford Gradient Spaces Lab.
- Jan 2026 Co-organizing the 3rd Workshop on Visual Concepts @ CVPR 2026.
- Apr 2025 Selected for the Google Gemini Academic Program.
- Apr 2025 Co-organizing the London GenAI Forum '25.
- Mar 2025 Awarded the Meshy AI Fellowship 2025 Finalist prize.
- Oct 2024 Presented my research at The Alan Turing Institute PhD day.
- Oct 2024 Started my PhD at UCL 🥳
Publications
Service
I review papers for the following conferences and journals:
- Computer Graphics SIGGRAPH · SIGGRAPH Asia · ACM Transactions on Graphics · Computer Graphics Forum
- Computer Vision CVPR · ICCV
- Machine Learning NeurIPS · ICLR · ICML · AAAI