Dwip Dalal

Hi, I am a third-year PhD Candidate at the University of Illinois Urbana‑Champaign, advised by Svetlana Lazebnik. I also closely collaborate with Unnat Jain and Heng Ji. I am working on post-training methods to build models for robotics that generalize. Some keywords to easily convey my research: MoT, VLAs, WAMs, VLMs, and multi‑modal representation learning.

Previously, I completed my B.Tech from Indian Institute of Technology, Gandhinagar, where I was awarded the Institute Gold medal for graduating at the top of my discipline.

I had great summers at Amazon (Ring AI, 2026) and Microsoft Research (2025), where I worked on long-video understanding with MLLMs and on vision language navigation with MLLM agents, respectively. Check out the projects: Repairable Uncertainty (Amazon) and AgentNav (Microsoft Research).

Internships

Amazon
Summer 2026

Microsoft Research
Summer & Fall 2025

University of British Columbia
Summer 2023

USC (AIISC)
2022‑2024

ISRO
Spring 2023

TCS Research
Winter & Spring 2023

EFICENS
Summer 2022

News

Jun 2026: 🎉🎉 Awarded the Dan Vivoli Endowed Fellowship by the ECE department at UIUC for the 2026–2027 academic year.
Jun 2026: 🎉🎉 Selected for Y Combinator's AI Startup School 2026.
May 2026: 🤘🤘 Summer internship at Amazon Ring AI (Sunnyvale), focusing on MLLM and video understanding.
Apr 2026: 🎉🎉 Passed my PhD Qualifying Exam — now a PhD Candidate! (Committee: Alexander Schwing, Huan Zhang, Mark Hasegawa-Johnson)
Apr 2026: 🎉🎉 Awarded the Qualcomm Graduate Award for the 2025–2026 academic year.
Apr 2026: 🎉🎉 Awarded a scholarship from Modal to support my research on VLAs.
Feb 2026: 🎉🎉 Our work on self-correcting MLLMs with attention-guided image warping was accepted for oral presentation at CSLSC 2026 (UIUC).
Jan 2026: 🎉🎉 First author paper accepted at ICLR 2026.
Jan 2026: 🎉🎉 Two first author papers accepted at EACL 2026, with one selected for oral presentation.
May 2025: 🤘🤘 Summer internship at Microsoft Research (Redmond), focusing on MLLM agents and reasoning.
Aug 2024: 🎉🎉 Awarded the Dilip and Sandhya Sarwate Graduate Fellowship at UIUC for the 2024‑2025 academic year.

Selected Publications

All Publications »

	Generalizable VLA Finetuning via Representation Anchoring and Language-Action Alignment Dwip Dalal, Shivansh Patel, Chahit Jain, Jeonghwan Kim, Utkarsh Mishra, Alex Baratian, Hyeonjeong Ha, Heng Ji, Svetlana Lazebnik, Unnat Jain *Under Review* Paper \| Code \| Project Page
Coming soon!	Repairable Uncertainty: Selective Temporal Reacquisition for Long-Video Understanding with MLLMs Dwip Dalal, Keval Doshi, Amar Kumar, Wei Wang, Sowndarya Sundar, Noranart Vesdapunt, Jim Thomas, Kah Kuen Fu, Svetlana Lazebnik *Under Review*
	Constructive Distortion: Improving MLLMs with Attention-Guided Image Warping Dwip Dalal, Gautam Vashishtha, Utkarsh Mishra, Jeonghwan Kim, Madhav Kanda, Hyeonjeong Ha, Svetlana Lazebnik, Heng Ji, Unnat Jain ICLR 2026 Paper \| Code \| Project Page
	Can MLLMs Find Their Way in a City? Exploring Emergent Navigation from Web-Scale Knowledge *Dwip Dalal,** Utkarsh Mishra, Narendra Ahuja, Nebojsa Jojic EACL 2026* (Oral) Paper \| arXiv \| Code \| Project Page
	Compositional Reasoning via Joint Image and Language Decomposition *Dwip Dalal,** Madhav Kanda, Zhenhailong Wang, Heng Ji, Unnat Jain EACL 2026* Paper \| Project Page
	Learning Robust Deep Visual Representations from EEG Brain Recordings Prajwal Singh, Dwip Dalal, Gautam Vashishtha, Shanmuganathan Raman, Krishna Prasad Miyapuram WACV 2024 Paper \| Code \| Project Page \| WACV Daily \| Best of WACV 2024
	Flow Symmetrization for Parameterized Constrained Diffeomorphisms *Dwip Dalal,** Aalok Gangopadhyay, Progyan Das, Shanmuganathan Raman Arxiv, 2024 Paper \| Code \| Project Page
	Single Image LDR to HDR Conversion Using Conditional Diffusion Dwip Dalal, Gautam Vashishtha, Prajwal Singh, Shanmuganathan Raman International Conference on Image Processing (ICIP'23) (Oral) Paper \| Project Page
	ODESolvers are also Wayfinders: Neural ODEs for Multi-Agent Pathplanning *Dwip Dalal,** Progyan Das, Anirban Dasgupta NeurIPS 2023 Workshop - Deep Learning and Differential Equations III* Paper \| Code
	FACTIFY-5WQA: 5W Aspect-based Fact Verification through Question Answering Anku Rani, SM Tonmoy, Dwip Dalal, Shreya Gautam, Megha C., Aman Chadha, Amit Sheth, Amitava Das ACL 2023 Paper \| Code & Dataset

Template Credits: Jon and Saurabh

	Generalizable VLA Finetuning via Representation Anchoring and Language-Action Alignment Dwip Dalal, Shivansh Patel, Chahit Jain, Jeonghwan Kim, Utkarsh Mishra, Alex Baratian, Hyeonjeong Ha, Heng Ji, Svetlana Lazebnik, Unnat Jain *Under Review* Paper \| Code \| Project Page
Coming soon!	Repairable Uncertainty: Selective Temporal Reacquisition for Long-Video Understanding with MLLMs Dwip Dalal, Keval Doshi, Amar Kumar, Wei Wang, Sowndarya Sundar, Noranart Vesdapunt, Jim Thomas, Kah Kuen Fu, Svetlana Lazebnik *Under Review*
	Constructive Distortion: Improving MLLMs with Attention-Guided Image Warping Dwip Dalal, Gautam Vashishtha, Utkarsh Mishra, Jeonghwan Kim, Madhav Kanda, Hyeonjeong Ha, Svetlana Lazebnik, Heng Ji, Unnat Jain ICLR 2026 Paper \| Code \| Project Page
	Can MLLMs Find Their Way in a City? Exploring Emergent Navigation from Web-Scale Knowledge *Dwip Dalal,** Utkarsh Mishra, Narendra Ahuja, Nebojsa Jojic EACL 2026* (Oral) Paper \| arXiv \| Code \| Project Page
	Compositional Reasoning via Joint Image and Language Decomposition *Dwip Dalal,** Madhav Kanda, Zhenhailong Wang, Heng Ji, Unnat Jain EACL 2026* Paper \| Project Page
	Learning Robust Deep Visual Representations from EEG Brain Recordings Prajwal Singh, Dwip Dalal, Gautam Vashishtha, Shanmuganathan Raman, Krishna Prasad Miyapuram WACV 2024 Paper \| Code \| Project Page \| WACV Daily \| Best of WACV 2024
	Flow Symmetrization for Parameterized Constrained Diffeomorphisms *Dwip Dalal,** Aalok Gangopadhyay, Progyan Das, Shanmuganathan Raman Arxiv, 2024 Paper \| Code \| Project Page
	Single Image LDR to HDR Conversion Using Conditional Diffusion Dwip Dalal, Gautam Vashishtha, Prajwal Singh, Shanmuganathan Raman International Conference on Image Processing (ICIP'23) (Oral) Paper \| Project Page
	ODESolvers are also Wayfinders: Neural ODEs for Multi-Agent Pathplanning *Dwip Dalal,** Progyan Das, Anirban Dasgupta NeurIPS 2023 Workshop - Deep Learning and Differential Equations III* Paper \| Code
	FACTIFY-5WQA: 5W Aspect-based Fact Verification through Question Answering Anku Rani, SM Tonmoy, Dwip Dalal, Shreya Gautam, Megha C., Aman Chadha, Amit Sheth, Amitava Das ACL 2023 Paper \| Code & Dataset

Dwip Dalal

Internships

News

Selected Publications

Generalizable VLA Finetuning via Representation Anchoring and Language-Action Alignment

Repairable Uncertainty: Selective Temporal Reacquisition for Long-Video Understanding with MLLMs

Constructive Distortion: Improving MLLMs with Attention-Guided Image Warping

Can MLLMs Find Their Way in a City? Exploring Emergent Navigation from Web-Scale Knowledge

Compositional Reasoning via Joint Image and Language Decomposition

Learning Robust Deep Visual Representations from EEG Brain Recordings

Flow Symmetrization for Parameterized Constrained Diffeomorphisms

Single Image LDR to HDR Conversion Using Conditional Diffusion

ODESolvers are also Wayfinders: Neural ODEs for Multi-Agent Pathplanning

FACTIFY-5WQA: 5W Aspect-based Fact Verification through Question Answering