Summary

Ph.D. candidate in Computer Application Technology at Fudan University, working on efficient and robust deep learning (data efficiency, model efficiency, OOD detection) and multi-agent LLMs for science.

Education

• Computer Application Technology

  2027 (expected)
  Fudan University
• Computer Science and Technology

  2021-06
  Fudan University

Work Experience

• LLM Algorithm Intern

  2026-06 -
  Research on agentic data processing for LLM training, building LLM-agent-based pipelines for data cleaning, synthesis, and quality control; exploration of data scaling laws, characterizing how data scale, mixture, and quality affect model performance.
• Research Intern

  2026-04 - 2026-05
  Led the design of a hierarchical benchmark and a multi-dimensional evaluation framework, and conducted a systematic survey of agent systems and materials benchmarks in the field; researched multi-agent LLM systems for automating Density Functional Theory (DFT) computation.
• LLM Pre-training Algorithm Intern

  2025-11 - 2026-04
  Investigated Scaling Laws and architecture optimization for MoE models, designing a pipeline mapping compute budgets to optimal architecture configurations; researched efficient and robust data-mixture algorithms using low-rank modeling of cross-domain interactions.
• Deep Learning Intern

  2020-03 - 2020-07
  Conducted deep-learning research for tumor precision medicine; designed and developed Graph Message Passing Neural Network (GMPNN) tools for molecular property analysis.