My name is Jingyan Shen. I am a first-year CS Ph.D. student at New York University (Courant Institute), advised by Prof. Matus Telgarsky and Prof. Pavel Izmailov. Previously, I earned my dual master degree from Tsinghua University and Columbia University. Prior to this, I completed my Bachelor’s degree at Wuhan University, majoring in Statistics.

I am broadly interested in machine learning and statistics. My current research studies when and why RL-based post-training improves reasoning in large language models. In particular, I am interested in:

• RL scaling and model priors: how the benefits and dynamics of RL post-training are shaped by what a model has already learned during pretraining or supervised fine-tuning.
• Weak-to-strong generalization: how RL can elicit stronger reasoning from limited, noisy, or weak supervision.
• Continual self-improvement: how LLMs can improve beyond fixed human-curated data through self-play, self-generated curricula, and verifier-guided exploration.

I am also fortunate to work with many great scholars and mentors, and I’m deeply grateful for their guidance.

🗞 News

🔬 Preprints & Workshops

(^†: equal contribution)

Preprint

Understanding Reasoning from Pretraining to Post-Training
Jingyan Shen^†, Ang Li^†, Salman Rahman, Yifan Sun, Micah Goldblum, Matus Telgarsky, Pavel Izmailov
Preprint. [Paper] [Code]

Preprint

When Can LLMs Learn to Reason with Weak Supervision?
Salman Rahman^†, Jingyan Shen^†, Anna Mordvina, Hamid Palangi, Saadia Gabriel, Pavel Izmailov
Preprint. [Paper] [Project Page]

NeurIPS 2025 Workshop

When Reasoning Meets Its Laws
Junyu Zhang^†, Yifan Sun^†, Tianang Leng^†, Jingyan Shen^†, Ziyin Liu, Paul Pu Liang, Huan Zhang
Efficient Reasoning Workshop at NeurIPS 2025 (Oral Presentation, Best Paper Nomination) [Paper] [Website]

🔬 Selected Publications

ICML 2026

Adversarially Robust Control of Conditional Value-at-Risk via Rockafellar-Uryasev Conformal Inference
Catherine Chen, Jingyan Shen, Zhun Deng, Lihua Lei
ICML 2026 [Paper]

EMNLP 2025

MiCRo: Mixture Modeling and Context-aware Routing for Personalized Preference Learning
Jingyan Shen^†, Jiarui Yao^†, Rui Yang^†, Yifan Sun, Feng Luo, Rui Pan, Tong Zhang, Han Zhao
EMNLP 2025 (Main) Outstanding Paper Award 🏆 [Paper]

NeurIPS 2025

Improving Data Efficiency for LLM Reinforcement Fine-tuning Through Difficulty-targeted Online Data Selection and Rollout Replay
Yifan Sun^†, Jingyan Shen^†, Yibin Wang^†, Tianyu Chen, Zhendong Wang, Mingyuan Zhou, Huan Zhang
NeurIPS 2025 [Paper]

ICLR 2025

TimeInf: Time Series Data Contribution via Influence Functions
Yizi Zhang^†, Jingyan Shen^†, Xiaoxue Xiong^†, Yongchan Kwon
ICLR 2025 [Paper] [Code]

🎖 Honors and Awards

• ICML 2026 Gold Reviewer, 2026
• Outstanding Paper Award, EMNLP 2025
• MacCracken Fellowship, New York University, 2025
• Outstanding Graduate Student (Top 1%), Tsinghua University, 2024
• Excellent Graduate Thesis Award, Tsinghua University, 2024
• Graduate Fellowship, Columbia University, 2023
• Outstanding Undergraduate Student, Wuhan University, 2021
• National Scholarship for Undergraduates, Ministry of Education of China, 2018

👾 Industry Experience

• 2024.02 - 2025.06, Full-time machine learning engineer at Pinterest

🐰 Miscs

I value the diversity and richness of life, and I hold a deep respect for beauty and purity in all their forms. Beyond research, I enjoy playing table tennis and tennis, playing drums, losing myself in a good book, or discovering new places through travel.