About

I am an AI Research Scientist at Meta Reality Labs. Previously, I was a research intern at Google AIR (affiliated with Google DeepMind) and Alibaba Damo Academy, focusing on on-device intelligence and LLM personalization. I earned my PhD in Computer Science from Worcester Polytechnic Institute, advised by Prof. Xiaozhong Liu.

My research focuses on Natural Language Processing, with particular emphasis on LLM personalization, memory mechanisms for large language models, LLM cascade optimization, and on-device intelligence. I am interested in building intelligent systems that understand, adapt to, and personalize for individual users.

LLM Personalization Memory Mechanism LLM Cascade On-device Intelligence Graph NLP

Experience

Meta Reality Labs

2025 – Present

AI Research Scientist

Worcester Polytechnic Institute

2021 – 2025

PhD in Computer Science · Advised by Prof. Xiaozhong Liu

Google AIR (DeepMind)

Jul – Oct 2024

Student Researcher · On-device Intelligence

Alibaba Damo Academy (Tongyi Lab)

May – Dec 2023

Research Intern

MIT

Previously

Research Intern

Publications

SaliMory: Orchestrating Cognitive Memory for Conversational Agents

Kai Zhang, et al. · arXiv 2026

Paper

Knowledge-based Citation Reasoning for Biomedical Domain

P Li, K Zhang, X Liu, X Zhang · Bioinformatics 2026

AssoMem: Scalable Memory QA with Multi-Signal Associative Retrieval

Kai Zhang, et al. · ICLR 2026

Paper Code

Privacy-preserved LLM Cascade via CoT-enhanced Policy Learning

Kai Zhang, et al. · WSDM 2026

Paper Code

KedRec-LM: A Knowledge-Distilled Explainable Drug Recommendation Large Language Model

Kai Zhang, et al. · SIGIR 2025 Workshop

Paper

Personalized LLM Response Generation with Parameterized Memory Injection

Kai Zhang, et al. · arXiv 2024

Paper Code

Knowledge Triplets Derivation from Scientific Publications via Dual-Graph Resonance

Kai Zhang, et al. · COLING 2024

Paper Code

LLM-based Medical Assistant Personalization with Short- and Long-Term Memory Coordination

Kai Zhang, et al. · NAACL 2024

Paper Code

Content- and Topology-Aware Representation Learning for Scientific Multi-Literature

Kai Zhang, et al. · EMNLP 2023

Paper Code

Chinese Agricultural Entity Relation Extraction via Deep Learning

Kai Zhang, et al. · ICIC 2019

Projects

AssoMem

Scalable Memory QA with Multi-Signal Associative Retrieval

11

MaLP

LLM-based Medical Assistant Personalization with Short- and Long-Term Memory Coordination

14

MiLP

Personalized LLM Response Generation with Parameterized Memory Injection

1

Privacy-preserving LLM Cascade

Source code for Privacy-preserving LLM Cascade

3

SMRC2

Self-supervised graph representation learning for breaking the boundary between text and graph

1

DGRN

Dual Graph based End model for SciIE

3

Contact

kzhang8@wpi.edu / kz8@iu.edu

Boston, MA