Yuanpu Cao
in 2023, and my B.Eng. in Software Engineering from Nankai University
in 2020, where I worked with Bloomberg
Jun–Sep 2026AI Research Intern
Developed topology-aware process supervision for reinforcement learning and iterative self-improvement methods for robust multi-hop reasoning in language models.
About Me
I am a fourth-year Ph.D. candidate in the College of Information Sciences and Technology at Penn State University, advised by Prof. Jinghui Chen. I received my M.S. in Computer Science from the Courant Institute of Mathematical Sciences at New York University in 2023, and my B.Eng. in Software Engineering from Nankai University in 2020, where I worked with Prof. Shenglin Zhang and Prof. Yongqian Sun.
Research Interests: My research focuses on language models, particularly post-training for reasoning and agentic capabilities, as well as model steering and reliability from a hidden representation perspective.
I am currently open to Research Scientist opportunities in industry, including internships and full-time roles. Please feel free to reach out!
Internship Experience
Amazon Web Services
May–Aug 2025Applied Scientist Intern · Agentic AI
Worked on efficient LLM agents through context compression and agent distillation.
Selected Publications
Full publications on Google Scholar
E indicates authors with equal contribution
Progress-Aware Distillation for Mitigating Stagnation in Small Language Model Agents
Restoring the Sweet Spot: Pass-Rate Weighted Self-Distillation for LLM Reasoning
ForecastCompass: Guiding Agentic Forecasting with Adaptive Factor Memory
From Narrative to Auditable Forecasts: A Structured Scaffold for Agentic Forecasting
Can Factual Opinions Be Edited (Manipulated) in Large Language Models?
TruthFlow: Truthful LLM Generation via Representation Flow Correction
Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization
Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
Stealthy and Persistent Unalignment on Large Language Models via Backdoor Injections
Preflect: From retrospective to prospective reflection in large language model agents
Towards robust multimodal large language models against jailbreak attacks
The Illusion of Reasoning: Exposing Evasive Data Contamination in LLMs via Zero-CoT Truncation
ICDAGENT: Empowering Agentic Large Language Models for Explainable Medical Coding
ParaBlock: Communication-Computation Parallel Block Coordinate Federated Learning for Large Language Models
Wordgame: Efficient & effective llm jailbreak via simultaneous obfuscation in query and response
Shadow-activated backdoor attacks on multimodal large language models
You Can't Steal Nothing: Mitigating Prompt Leakages in LLMs via System Vectors
Advi2i: Adversarial image attack on image-to-image diffusion models
Guarddoor: Safeguarding against malicious diffusion editing via protective backdoors
Phi: Preference Hijacking in Multi-modal Large Language Models at Inference Time
Tackling the data heterogeneity in asynchronous federated learning with cached update calibration
Adversarially robust industrial anomaly detection through diffusion model
Federated Learning with Projected Trajectory Regularization
CTF: Anomaly detection in high-dimensional time series with coarse-to-fine model transfer
RLCard: A Toolkit for Reinforcement Learning in Card Games
CoFlux: Robustly Correlating KPIs by Fluctuations for Service Troubleshooting
Education
Penn State University
Aug. 2023–PresentPh.D. in Information Sciences and Technology
New York University
Aug. 2021–May 2023M.S. in Computer Science · Courant Institute of Mathematical Sciences
Nankai University
Sep. 2016–Jul. 2020B.Eng. in Software Engineering
Honors and Awards
Herbert J. Bailey Graduate Fellowship
2026–2027Selected as the sole nominee from the College of Information Sciences and Technology at Penn State University.
NeurIPS 2024 Scholar Award
2024PSU IST Student Travel Award
2024School Scholarship for Excellent Academic Performance
2018Nankai University
Academic Service
Program Committee/Reviewer
- Neural Information Processing Systems (NeurIPS)
- International Conference on Machine Learning (ICML)
- International Conference on Learning Representations (ICLR)
- Association for the Advancement of Artificial Intelligence (AAAI)
- ACL Rolling Review (ARR)
- Transactions on Machine Learning Research (TMLR)