I am Yuelin Hu, a Research Master student at Shanghai Jiao Tong University (SJTU), specializing in large language models, post-training alignment, and agent-based systems. My academic training combines a strong theoretical foundation with intensive engineering practice, covering machine learning, deep learning, optimization, and AI systems.

I have conducted research and engineering work at multiple leading industrial research labs, including Xiaohongshu Hi Lab, Ant Group MYbank AI Department, and Microsoft Research Asia (MSRA). My experience spans large-scale supervised fine-tuning (SFT), reinforcement learning for LLMs (GRPO, DPO, PPO variants), data quality pipelines, evaluation infrastructure, and end-to-end model deployment on thousand-GPU clusters.

On the research side, I am the first author of multiple papers submitted to or accepted by top-tier conferences such as AAAI, ACL, ICASSP, and ISCAS. My work focuses on stabilizing and improving LLM alignment through principled training control, credit assignment, and agent-based reasoning, including MCTS-enhanced GRPO, adaptive SFT–RL mixing strategies, and domain-aware web agents.

Overall, I am deeply motivated by building reliable, scalable, and well-evaluated intelligent systems, and I aim to continue contributing to both academic research and real-world LLM applications.