CV

Education

• Ph.D in Information Science, Drexel University, 2026 (expected)
• M.S. in Information Science, University of North Carolina at Chapel Hill, 2021
• B.S. in Management Information System, Dalian University of Technology, 2019

Work experience

NLP Research Intern

• Samsung Research America, Mountain View, California (September 2025 – now)
  • Working on improving QA Grounding with layout information.

Applied Scientist Intern

• Amazon, Seattle, Washington (June 2025 – September 2025)
  • Designed and implemented a novel framework that generates precise product category definitions through reinforcement learning from verbal feedback, replacing costly manual definition writing
  • Built a modular system leveraging Claude Sonnet 3.5 and Claude Haiku 3.5, including components for the actor, evaluators, self-reflection, and a memory module to store historical reflections
  • Designed a sampling model to extract representative product examples from millions of listings, enabling scalable learning
  • Achieved 97.5% classification accuracy (vs. 94% human-crafted definitions) while generating shorter, clearer outputs
  • Reduced category definition creation from weeks to hours, accelerating business response to new product categories and emerging market demands

Research Scientist Intern

• Ping An Technology Research Lab, Palo Alto, California (March 2025 – June 2025)
  • Designed and optimized both ToB and ToC medical VLM, supporting real-world deployment in general practice settings
  • Fine-tuned 32B, 72B Qwen vision-language large models (VLMs) on 400K multi-turn medical consultation dialogues using LoRA and DeepSpeed on 8×A800 GPUs
  • Developed a RAG (Retrieval-Augmented Generation) pipeline with a knowledge graph of rare diseases and treatments, improving diagnostic accuracy from 82% to 90% while reducing hallucinations
  • Quantized Qwen 72B VLM with GPTQ, decreasing model size by 69.65%
  • Deployed the quantized model using vLLM for inference acceleration, reducing response latency by 50–66% compared to baseline

Skills

• Model Training
  • TensorFlow, PyTorch
  • SFT, RLHF, PPO, QLoRA
  • Distributed RPC (Data/Model Parallel)
  • Quantization Aware Training
• Model Deployment
  • Docker, Kubernetes, Tensorflow Lite
  • React.js, Django, Node.js, Spring Cloud
  • AWS EC2
• Programming Languages
  • Python (transformers, opencv, nltk, sklearn, scipy)
  • Java, JavaScript, C, C#
  • Shell/Scripting
  • SQL/NoSQL

Service

• Reviewer for the AAAI Conference on Artificial Intelligence (AAAI), 2025
• Reviewer for the Conference on Empirical Methods in Natural Language Processing (EMNLP), 2025
• Reviewer for the 63rd Annual Meetings of the Association for Computing Linguistics (ACL), 2025
• Reviewer for International Conference on Computational Linguistics (COLING), 2024

Academic Activities

• Oral Presentation at COLING, January 2025
• Oral Presentation at IJCAI XAI workshop, June 2024
• Poster Presentation at AAAI Doctoral Consortium, February 2025
• Poster Presentation at AAAS, February 2024
• Poster Presentation at AAAI IAAI session, Washington D.C., 2023

Awards & Scholarships

• Phoebe W. Haas Endowed Fellowship for Women Doctoral Students, 2023
• Honorable Mention (2nd place) in Mathematical Contest in Modeling (MCM), USA, 2017