Yutong Song

Computer Science PhD candidate @ UCI.

About me 👋

This is Yutong Song, a forth-year PhD student at the University of California, Irvine and I will graduate in 2026. I am now living in Irvine with my cute cat Hula. I am working in Health SciTech Group, under the mentorship of Prof.Amir M. Rahmani and Prof.Nikil Dutt.

News 🥳

Selected Publications 🚀

Google Scholar
Overview of Do Implicit Personalization and Explicit Styles Conflict? PsPLUG: A Lightweight Plug-in for Balancing Personalization and Style in Customized LLMs

EMNLP 2026

Do Implicit Personalization and Explicit Styles Conflict? PsPLUG: A Lightweight Plug-in for Balancing Personalization and Style in Customized LLMs

Yutong Song, Jiang Wu, Shaofan Yuan, Chengze Shen, Jian Wang, Yu Wang, Nikil Dutt, Amir M. Rahmani

PsPLUG learns a user-specific residual after accounting for the requested style, preserving implicit user preferences while offering precise control over personalization strength and style adherence.

Overview of CARD: Cluster-level Adaptation with Reward-guided Decoding for Personalized Text Generation

EMNLP 2026 Findings

CARD: Cluster-level Adaptation with Reward-guided Decoding for Personalized Text Generation

Yutong Song, Jiang Wu, Weijia Zhang, Chengze Shen, Shaofan Yuan, Weitao Lu, Jian Wang, Yu Wang, Nikil Dutt, Amir M. Rahmani

CARD is a hierarchical personalization text generation framework that combines group-level adaptation with user-specific decoding to generate text better aligned with individual preferences.

Overview of AP-GRPO: Anchor-Gated Phonetic Alignment with Policy Optimization for Pathological Speech Reconstruction

EMNLP 2026

AP-GRPO: Anchor-Gated Phonetic Alignment with Policy Optimization for Pathological Speech Reconstruction

Pengfei Zhang, Hoang H. Nguyen, Yutong Song, Wenjun Huang, Tahmid Imtiaz Imu, Henry Peng Zou, Jiang Wu, Honghui Xu, Amir M. Rahmani

AP-GRPO improves pathological speech reconstruction with anchor-gated and inter-anchor phonetic rewards that preserve reliable speech evidence while recovering corrupted content.

Overview of DemMA: Dementia Multi-Turn Dialogue Agent with Expert-Guided Reasoning and Action Simulation

ACL 2026 Findings

DemMA: Dementia Multi-Turn Dialogue Agent with Expert-Guided Reasoning and Action Simulation

Yutong Song, Jiang Wu, Kazi Sharif, Honghui Xu, Nikil Dutt, Amir Rahmani

DemMA is an expert-guided dementia dialogue agent that jointly models cognitive impairment, emotional dynamics, and nonverbal behaviors for high-fidelity multi-turn patient simulation.

Overview of MedSpeak: A Knowledge Graph-Aided ASR Error Correction Framework for Spoken Medical QA

ICASSP 2026

MedSpeak: A Knowledge Graph-Aided ASR Error Correction Framework for Spoken Medical QA

Yutong Song, Shiva Shrestha, Chenhan Lyu, Elahe Khatibi, Pengfei Zhang, Honghui Xu, Nikil Dutt, Amir Rahmani

MedSpeak uses semantic relationships and phonetic information from a medical knowledge graph, together with LLM reasoning, to correct noisy ASR transcripts and improve spoken medical question answering.

Working Experience 🔑

Zillow Group

Applied Scientist Intern

  • Developed HIPPO-Agent, a training-free memory controller that wraps a fixed base agent with retrieve, explore, evaluate, and write operations, using L1 for answer-blind failure reflections and L2 for compact, transferable strategies.
  • Built the Tour Day Co-Pilot hackathon MVP with AgentScope, implementing multimodal onsite capture and image captioning to support AI-generated, cross-home decision debriefs and a consent-gated personalization memory loop.

TikTok

Research Scientist Intern

  • Contributed to MCM, a foundation user model for content creators, implementing personalized in-app messaging and inspiration discovery for the Creator Search Insights (CSI) platform.
  • Proposed a personalized text-to-text generation framework that injects user-specific preferences during decoding while leveraging a fine-tuned foundation model.

Education 🎓

University of California, Irvine

PhD student in Computer Science

University of California, Irvine

PhD student in Computational Science

University of Electronic Science and Technology of China

B.Eng. in Communication Engineering

Awards and Honors 🏆