Jiaju Han

I am currently pursuing a second bachelor's degree in Data Science and Big Data Technology at China University of Petroleum (Beijing), after completing a B.A. in English at the same university. I am working on research projects related to vision-language models, large language models, and multimodal AI, with current experience in VLM robustness evaluation, LLM-based generation systems, and retrieval-augmented reasoning.

My research interests include computer vision, large language models, retrieval-augmented generation, artificial intelligence, and embodied AI. I am especially interested in various applications of large language models.

I am looking for 2027 PhD opportunities, research internships, visiting student positions, and research assistant opportunities, especially in LLM applications and RAG.

Updates

  • May 2026 Submitted EXPOSING VULNERABILITIES IN VISIBLE-INFRARED VLMs: A UNIFIED GEOMETRIC ADVERSARIAL FRAMEWORK WITH CROSS-TASK TRANSFERABILITY to NeurIPS.
  • May 2026 Submitted FROM CLOUDS TO HALLUCINATIONS: ATMOSPHERIC RETRIEVAL HIJACKING IN REMOTE SENSING VISION-LANGUAGE RAG to NeurIPS.
  • Mar 2026 Submitted XSPA: CRAFTING IMPERCEPTIBLE X-SHAPED SPARSE ADVERSARIAL PERTURBATIONS FOR TRANSFERABLE ATTACKS ON VLMs to ACM Multimedia.
  • Mar 2026 Submitted WHEN SURFACES LIE: EXPLOITING WRINKLE-INDUCED ATTENTION SHIFT TO ATTACK VISION-LANGUAGE MODELS to ACM Multimedia.

Publications

CFGPatch teaser figure

NeurIPS Submission · arXiv Preprint

EXPOSING VULNERABILITIES IN VISIBLE-INFRARED VLMs: A UNIFIED GEOMETRIC ADVERSARIAL FRAMEWORK WITH CROSS-TASK TRANSFERABILITY

Xiang Chen, Yuxian Dong, Chao Li, Chengyin Hu, Jiaju Han, Fengyu Zhang, Yiwei Wei, Jiahuan Long, Jiujiang Guo

[arXiv]

CloudWeb teaser figure

NeurIPS Submission · arXiv Submission Preview

FROM CLOUDS TO HALLUCINATIONS: ATMOSPHERIC RETRIEVAL HIJACKING IN REMOTE SENSING VISION-LANGUAGE RAG

Jiaju Han, Chao Li, Chengyin Hu, Qike Zhang, Xuemeng Sun, Xin Wang, Fengyu Zhang, Xiang Chen, Yiwei Wei, Jiahuan Long, Jiujiang Guo

[arXiv]

XSPA teaser figure

ACM MM Submission · arXiv Preprint

XSPA: CRAFTING IMPERCEPTIBLE X-SHAPED SPARSE ADVERSARIAL PERTURBATIONS FOR TRANSFERABLE ATTACKS ON VLMs

Chengyin Hu, Jiaju Han, Xuemeng Sun, Qike Zhang, Yiwei Wei, Ang Li, Chunlei Meng, Xiang Chen, Jiahuan Long

[arXiv]

When Surfaces Lie teaser figure

ACM MM Submission · arXiv Preprint

WHEN SURFACES LIE: EXPLOITING WRINKLE-INDUCED ATTENTION SHIFT TO ATTACK VISION-LANGUAGE MODELS

Chengyin Hu, Xuemeng Sun, Jiaju Han, Qike Zhang, Xiang Chen, Xin Wang, Yiwei Wei, Jiahua Long

[arXiv]

Projects

LLM-Based Two-Stage Credit Rating Explanation Generation

Led an LLM-based system covering rating prediction, explanation generation, LLM-as-judge evaluation, semantic feedback, and DPO signal construction. The project explores how LLMs can generate faithful and decision-relevant explanations.

ScholarRAG-ZH

Designed a citation-grounded QA benchmark and RAG evaluation toolkit for Chinese academic documents, comparing lexical retrieval, BM25, dense retrieval, cross-encoder reranking, and structure-aware retrieval under draft and noisy PDF settings.

Education

China University of Petroleum (Beijing) B.Eng. in Data Science and Big Data Technology, 2025 - 2027
China University of Petroleum (Beijing) B.A. in English, 2021 - 2025

Contact

I am actively seeking 2027 PhD opportunities, research internships, visiting student positions, and research assistant opportunities, especially in LLM applications and RAG.

Email: 1404293404 [at] qq [dot] com