Yi Yuan 袁懿
MurmanskY

Ph.D. student (integrated M.S.–Ph.D.) @ NISL Lab, Fudan University. Working on LLM-generated text detection, AIGC forensics & LLM security. I turn ideas into code. 📷 Photography · 🏎️ Petrolhead.

langPython dlPyTorch alsoC++ · Java · SQL · Solidity fieldAIGC Forensics

# About

Hi 👋 I'm a Ph.D. student in the integrated M.S.–Ph.D. program at Fudan University, working with the NISL Lab. Before that, I got my B.Eng. in Computer Science (Excellence Program) from Nankai University, and spent a year as a research intern at Tsinghua's Mengshi Lab.

My work centers on trustworthy & secure AI — detecting LLM-generated text, AIGC forensics, and poking at how large models can be misused (e.g., as carriers for hidden payloads). Research interests: llm-text-detection aigc-forensics llm-security steganography multi-agent

# News

  • 2026.10dualrankDualRank advanced to the second round of AAAI 2027 review 🚀
  • 2026asoc-acceptFirst-author paper Industrial Large Models as Carriers for Malicious Payloads accepted by Applied Soft Computing
  • 2026yuanqi-p2Kicked off Yuanqi Phase II as team lead — multi-agent evolution & long-term memory
  • 2025.10awardNamed Outstanding Student of Fudan University
  • 2025.04awardFirst-Class Academic Scholarship, Fudan University
  • 2024v2x-surveyV2X survey accepted by Automotive Innovation (I wrote Chapters 5–6)

# Publications

DualRank: Cross-Model Ranking Consistency for Zero-Shot LLM-Generated Text Detection

First author (Yi Yuan et al.). Zero-shot detection via cross-model token-ranking consistency — no training on generator outputs needed.

AAAI 2027 CCF-A Under review · R2
Industrial Large Models as Carriers for Malicious Payloads: A Fast and Robust Approach

First author (Yi Yuan et al.). Industrial-scale LLMs can act as carriers for hidden malicious payloads — a fast, robust construction with supply-chain security implications.

Applied Soft Computing Accepted
Vehicle-to-Everything Communication in Intelligent Connected Vehicles: A Survey and Taxonomy

Contributing author (Mengshi Lab, Tsinghua). Wrote Chapters 5–6; benchmarked 10+ autonomous-driving datasets and their metrics.

Automotive Innovation Accepted
OCID

Yi Yuan et al. Manuscript in preparation — details coming soon.

WIP In preparation

# Research & Experience

NISL Lab, Fudan University

Graduate researcher 2024.04 — now
  • Yuanqi Phase II · team lead — industry–university collaboration on multi-agent evolution optimization & long-term memory
  • Yuanqi Phase I · core member — LLM security & generated-text detection; concluded
  • Project 131 · team lead & core contributor — LLM security & steganography; passed final acceptance
  • Model training & fine-tuning across vision tasks (face recognition, detection, segmentation); front-end work for lab systems

Mengshi Lab, Tsinghua University

Research intern · School of Vehicle and Mobility 2023.06 — 2024.05
  • Co-wrote the V2X survey (Chapters 5–6), accepted by Automotive Innovation
  • Curated & compared 10+ autonomous-driving training datasets and evaluation metrics

# Teaching

TA — Undergraduate AI Course

Fudan University 2025
  • Revised slides/handouts, guided labs, designed & graded assignments and the final project
  • From environment setup to linear nets, MLPs and CNNs, ending with a hands-on capstone

Volunteer Python Instructor — "Into Campus"

Fudan Science Park Primary School 2025 Spring
  • Taught Python basics (syntax, functions, modules) to third-graders with interactive, project-based lessons

# Education

  • 2024–2029Fudan University — Ph.D. in Electronic Information (integrated M.S.–Ph.D.), Shanghai
  • 2019–2023Nankai University — B.Eng. in Computer Science, Excellence Program · GPA 85.4/100

# Honors

  • 2025.10Outstanding Student, Fudan University
  • 2025.04First-Class Academic Scholarship, Fudan University
  • 2021.03National-Level Project, National College Student Innovation & Entrepreneurship Competition (No. 202110055019)
  • 2020.12Innovation Scholarship, Nankai University