Agam Pandey
Joint Secretary, DSG IIT Roorkee
Education
-
B.Tech, Civil Engineering, IIT Roorkee (2022–2026)
Research Interests
- Multimodal Models
- Adversarial ML
- Causal Reasoning in VLMs
- Graph Neural Networks
- Multi Agent systems
- Computer Vision for Autonomous vehicles
- RAG, Knowledge Graphs for V/LLMs
Publications
- MLRC 2025 [Best Paper Award], TMLR: A. Mittal, A., Pandey, A., Tiwari, A., Jindal, S., Swain, S., (2025). Revisiting CroPA: A Reproducibility Study and Enhancements for Cross-Prompt Adversarial Transferability in VLMs.
- NeurIPS Reliable ML Workshop: A., Pandey, A.Mittal, A., Tiwari, A., Jindal, S., Swain, S.,, CroPA++: Exposing Vulnerabilities in Vision Language Models and Enhancing Adversarial Transferability of Cross-Prompt Attacks
- AAAI 2026 SAPP: TransPatch: Learning a Universal Adversarial Patch for ViT–CNN Cross-Architecture Transfer in Semantic Segmentation (Student Abstract) A.Pandey, S. Goyal, A. Tomar, A. Aggarwal, A. Tiwari.
- AAAI 2026 SAPP: Zero-Shot Vision Language Reasoning via Dual-layer Scene Graph Chain of Thoughts (Student Abstract), A. Pandey, P. Kapoor, Y. Bansal
Projects
- Knowledge Graph Embeddings for VLMs
Analysed impact of vector and graph embeddings on retrieval generation of VLMs on downstream tasks. - FeedCode
FeedCode is an intelligent feedback tool that provides personalized code reviews based on your coding style and activity.
Follow
Portfolio |
|
