Hi, I am Utkarsh Mishra, an MS student in Computer Engineering at Texas A&M University. I am currently working as an AI/ML/CV Algorithms Intern at Applied Materials. My research interests lie in Large Language Models
(LLMs), Multimodal LLMs, vision foundation models, agentic AI systems, reinforcement learning from human feedback (RLHF),
vision-language understanding, and robotics.
I graduated with a B.Tech in Electrical Engineering and a minor in Computer Science from IIT Gandhinagar. Previously, I worked as a System
Analyst at HDFC Bank.
I'm passionate about building intelligent systems that can reason, plan, and interact with the world
through multimodal understanding.
Let's connect to discuss innovative ideas in these spaces!
Industrial and Research Experience
AI/ML/CV Intern @ Applied Materials May 2026 - Present
Research Engineer @ TAMU Oct 2025 - May 2026
Research Collaborator @ UIUC Jan 2025 - Present
System Analyst @ HDFC Bank Jul 2024 - Aug 2025
Research Assistant @ IIT GN Aug 2023 - Feb 2025
News & Achievements
Aug 2026: Internship at Applied Materials extended through December 2026.
May 2026: Joined Applied Materials as an AI/ML/CV Algorithms Intern for the summer.
Jan 2026: Paper on Multimodal Large Language Models (MLLMs) accepted at International
Conference on Learning Representations (ICLR) 2026.
Jan 2026: Paper on City Navigation using MLLMs accepted at European Chapter of the
Association for Computational Linguistics (EACL) 2026 with Oral
Presentation.
Oct 2025: Started as Research Engineer at Texas A&M University (Dept. of AGLS).
Aug 2025: Started MS in Computer Engineering at Texas A&M University.
May 2025: Paper on 3D Point Cloud Denoising using Diffusion Transformers
accepted at International Conference on Image Processing (ICIP) 2025.
Mar 2025: Received admit from Texas A&M University with ECEN Merit Scholarship.
Jun 2024: Graduated from IIT Gandhinagar.
Publications
Generalizable VLA Finetuning via Representation Anchoring and Language-Action Alignment
Dwip Dalal, Shivansh Patel, Chahit Jain, Jeonghwan Kim, Utkarsh Mishra, Yuchen Song, Alex Baratian, Hyeonjeong Ha,
Heng Ji, Svetlana Lazebnik*, Unnat Jain*
Under Review Paper |
Code |
Project Page
Can MLLMs Find Their Way in a City? Exploring Emergent Navigation from Web-Scale Knowledge
Constructive Distortion: Improving MLLMs with Attention-Guided Image Warping
Dwip Dalal, G. Vashishtha, Utkarsh Mishra, J. Kim, M. Kanda, H. Ha, Svetlana
Lazebnik, Heng Ji, Unnat Jain
ICLR 2026 Paper |
Project Page |
Animated Presentation
Transformer Augmented Multi-Resolution Hash Encoding in Diffusion Model for
3D Point Cloud Denoising
A novel approach combining transformer architectures with multi-resolution hash encoding in
diffusion models for effective 3D point cloud denoising.
Projects
Two Maps of the Same Territory: Citation Structure vs. Semantic Content Spring 2026 - CSCE 676 Data Mining & Analysis, Texas A&M Code
|
Video
|
Animated Presentation
Built a graph and text mining pipeline over the ogbn-arxiv citation graph (169K CS papers) to measure
where citation structure and semantic content disagree, comparing Louvain citation communities with
k-means clusters of title-and-abstract embeddings. Found 145 high-PageRank 'Import Papers', Adam among them, whose citers sit far from them in semantic content, the way code imports a library, and a field-wide drift toward that pattern from 2010 to 2020.
Proposed heterogeneous attention for PINNsFormer, a physics-informed transformer PDE solver: each
attention head scores query-key pairs with its own nonlinear kernel (identity, ReLU, square, tanh or a
learned mix) instead of the shared dot product, aimed at a PDE's linear, quadratic and cubic terms. Slightly improved accuracy on Burgers over the unmodified PINNsFormer.
VGAP: Vision-Gated Action Planning for Web Agents Fall 2025 - CSCE 642 Deep Reinforcement Learning, Texas A&M Code
Developed a visual grounding module for autonomous web agents using Direct Preference Optimization
(DPO) to train a Qwen2-0.5B model for predicting optimal screenshot crop regions. Integrated into
SeeAct framework with LiteLLM support for GPT-4o, Claude, Gemini, LLaVA.
Analyzing Physics Understanding in Video Generation Models Fall 2025 - CSCE 689 Vision Foundation Models, Texas A&M Demo Page
Evaluated whether video generation models (Sora, Veo3, Grok, Wan2.1) can produce physically
realistic data for robotics training. Designed physics benchmark suite testing gravity, buoyancy,
and collision dynamics with SAM2-based kinematics analysis.
Denoising Gaussian Splatting for 3D Scene Reconstruction Guide: Prof. Ravi Hegde
Aug 2024 - Dec 2024 Code
|
Animated Presentation
Extended 3D Gaussian Splatting with denoising techniques (DBSCAN on the input point cloud, point-wise distance pairing to prune isolated Gaussians during training) and novel regularization techniques to reduce visual artifacts when using low
resolution wide angle images.
GAN Inversion for Latent Space Analysis Guide: Prof. Anirban Dasgupta
Jan 2024 - May 2024 Code
Used GAN inversion on StyleGAN to analyze effect of object rotation on latent representation and
generate novel views through latent space manipulation.
Human Pose Classification using Spatial-Temporal GNNs Guide: Prof. Ravi Hegde
Jan 2024 - May 2024 Code
|
Poster
Combined Graph Convolutional Networks with LSTM to model spatial relationships and temporal dynamics
for pose classification using OpenPose and AlphaPose.
Synthetic Data Generation for Machine Learning Guide: Prof. Shanmuganathan Raman
Aug 2023 - May 2024 Code
|
Poster
Generated high-quality synthetic images using StyleGAN-XL and Stable Diffusion for CIFAR-10 dataset.
Analyzed effects on classifier accuracy with varying synthetic to real data ratios.
Wearable Device for Real Time Sign Language Recognition Guide: Prof. Jhuma Saha
Apr 2023 Code
|
Animated Presentation
Developed wearable device using STM32 Nucleo microcontroller and flex sensors. Implemented USB CDC
protocol with Scikit-learn MLP classifier for gesture detection.