About me

I am a senior research engineer with Ant Group, where I currently work on a variety of LLM alignment and RL.

Before this, I worked in IBM Research AI as a research intern and collaborated with Pin-Yu Chen, Payel Das, Songtao Lu, Xiaodong Cui and many other talented researchers at IBM. My research in IBM focused on LLM alignment and offline RL. Meanwhile, I did my Ph.D. under the supervision of Dr. Tianyi Chen (now in Cornell Tech) in RPI. I was fortunate to join Dr. Tianyi Chen’s group as the first Ph.D. student. My Ph.D. research focused on optimization and reinforcement learning.

News and highlights

• [Sep. 2025] New paper on LLM-RL: AEnt. Its asynchronous implementation is incorporated in the highly scalable RL framework AReaL.
  • On Entropy Control in LLM-RL Algorithms [code]
• [Mar. 2025] I am excited to join Ant Group via its research talent program Ant Star.
• [Feb. 2025] Our extended study of the ICML 2024 paper is dual accepted in JMLR.
  • ArXiv: Principled Penalty-based Methods for Bilevel Reinforcement Learning and RLHF
• [Jan. 2025] Our paper is accepted in ICLR 2025:
  • SEAL: Safety-enhanced Aligned LLM Fine-tuning via Bilevel Data Selection
• [Dec. 2024] Our extended study of the ICML 2023 paper has been dual accepted in Mathematical Programming.
  • On Penalty-based Bilevel Gradient Descent Method
• [Oct. 2024] New paper on improved LLM alignment framwork:
  • Mitigating Forgetting in LLM Supervised Fine-Tuning and Preference Learning

Selected works

• SEAL: Safety-enhanced Aligned LLM Fine-tuning via Bilevel Data Selection
  Han Shen, Pin-Yu Chen, Payel Das, Tianyi Chen
  ICLR 2025. [arxiv]

• Mitigating Forgetting in LLM Supervised Fine-Tuning and Preference Learning
  Heshan Fernando*, Han Shen*, Parikshit Ram, Yi Zhou, Horst Samulowitz, Nathalie Baracaldo, Tianyi Chen
  *equal contribution, preprint. [arxiv]

• Principled Penalty-based Methods for Bilevel Reinforcement Learning and RLHF
  Han Shen, Zhuoran Yang, Tianyi Chen
  ICML 2024, extended work in JMLR [arxiv]

• Mitigating Gradient Bias in Multi-objective Learning: A Provably Convergent Approach
  Heshan D. Fernando, Han Shen, Miao Liu, Subhajit Chaudhury, Keerthiram Murugesan, Tianyi Chen
  ICLR 2023 oral. [arxiv]

• On Penalty-based Bilevel Gradient Descent Method
  Han Shen, Quan Xiao, Tianyi Chen
  ICML 2023, extended work in Mathematical Programming. [MAPR]

Industry experiences

Ant Group. (CN) Present

• Senior research engineer, joined via Ant Star talent program.

IBM Research AI. (US) 05.2024 - 08.2024

• Research intern, mentored by Dr. Pin-Yu Chen and managed by Dr. Payel Das.

IBM Research AI. (US) 05.2021 - 08.2021

• Research intern, mentored by Dr. Songtao Lu and Dr. Xiaodong Cui.

Services

Reviewer for NeurIPS, ICML, ICLR, AISTATS, AAAI and IEEE Transactions on Signal Processing (TSP).