I am a Master’s student in Machine Learning at Carnegie Mellon University (CMU). I received my B.S. in Data Science from the University of California, San Diego (UCSD) and the University of California, Santa Barbara (UCSB) (GPA: 3.96/4.0, Provost Honor, Top 2%).

My research interests include LLM systems, AI agents, explainable AI, and machine learning. I have experience building large-scale AI systems, developing research agents, and optimizing language model training and inference pipelines. I am currently a Research Assistant at UCSD MixLab x Meta AI, working on AI research agents and personalization, advised by Prof. Zhiting Hu, Prof. Hao Zhang, and Dr. Zhen Wang.

🔥 News

2026.02: 🎉🎉 Paper “FIRE-Bench” submitted to ICML 2026 (Under Review).
2025.11: 🎉🎉 Paper “DeepPersona” accepted as Spotlight at NeurIPS 2025 LAW Workshop.
2025.01: 🎉🎉 Paper “Exploration and practice of human-machine trustworthy mechanism in XAI” published on Big Data Research.

📝 Publications

FIRE-Bench: Evaluating Research Agents on the Rediscovery of Scientific Insights (ICML 2026 Under Review) Zhen Wang*, Fan Bai*, Zhongyan Luo*, Jinyan Su, Kaiser Sun, Weiqi Liu, Albert Chen, Jieyuan Liu, Kun Zhou, Claire Cardie, Mark Dredze, Eric P. Xing, Zhiting Hu
DeepPersona: A Generative Engine for Scaling Deep Synthetic Personas (NeurIPS 2025 LAW Spotlight) Zhen Wang*, Yufan Zhou*, Zhongyan Luo, Lyumanshan Ye, Adam Wood, Man Yao, Saab Mansour, Luoshang Pan
Exploration and practice of human-machine trustworthy mechanism in XAI (Published on Big Data Research) Zhongyan Luo, Zhengxun Xia, Jianfei Tang, Yifan Yang, Hongshan Yang, Haohua Li, Yan Zhang
(Patent) SQL generation method, device, equipment and medium based on background knowledge enhancement
(Patent) Question answering method and apparatus based on large language model, electronic device, and storage medium
(Patent) Information classification method, device, equipment and storage medium
(Patent) Method and device for query processing of label data, computer equipment and medium

🎖 Honors and Awards

2022 - 2024 Provost Honor, University of California, Santa Barbara (Top 2%)
2022 - 2024 GPA: 3.96/4.0

📖 Educations

2026.09 - 2028.06, M.S. in Machine Learning, Carnegie Mellon University (CMU), Pittsburgh, PA
2024.09 - 2026.06, B.S. in Data Science, University of California, San Diego (UCSD), San Diego, CA
2022.09 - 2024.06, B.S. in Data Science, University of California, Santa Barbara (UCSB), Santa Barbara, CA

💻 Experience

2025.04 - Present, Research Assistant, UCSD MixLab x Meta AI
- Developed AI Research Agent using LangGraph, integrating brainstorming, multi-agent discussion and runtime tools.
- Benchmarked scientific discovery ability of AI Research Agents by building a containerized environment using Docker.
- Built large scale human-attribute taxonomy and synthesized MB-level profiles, promoting better personalization in AI.
- Constructed a meta-learning framework for self-evolving agent skills, achieving 40% accuracy increase on SkillsBench.
2024.06 - 2024.09, AI Research Intern, Transwarp
- Implemented LoRA-based fine-tuning pipeline in PyTorch improving downstream tasks accuracy by around 15%.
- Established automated evaluation pipeline for 20+ Text2SQL benchmarks via vLLM, LLM APIs and Python SQLite.
- Built finance-oriented RAG system that integrates 200+ multimodal documents, achieving 30% accuracy increase.
- Built an XAI (Explainable AI) module through a single interface, making model behavior easier to audit and debug.
2023.06 - 2023.09, Software Engineer Intern, Transwarp
- Contributed to Kubernetes-based LLM training platform; used TorchX and Volcano to optimize resource allocation.
- Integrated Prometheus and Grafana dashboards for training & inference workload, reducing debugging time by 50%.
- Built a security gateway for AI applications and deployed via CI/CD pipelines, reducing information leakage by 98%.
- Automated microservice releases using Docker and Kubernetes deployment workflow for repeatable rollouts & scaling.

🛠 Projects