👋 About me

Hi, I’m Zhenqing Ling (凌振清), a researcher at Tongyi Lab (Token Foundry), Alibaba Group. I joined in July 2026, following a research internship at Tongyi from April 2025. I am mentored by Daoyuan Chen and Yaliang Li.

I received my master’s degree from Sun Yat-sen University in 2026, advised by Prof. Ying Shen. Before that, I received my bachelor’s degree from the School of Mathematics at Sun Yat-sen University in 2023.

My current research centers on data-centric AI for foundation models and agents. I explore data infrastructure for self-improving LLMs and agents, focusing on:

• Data–model co-evolution. Understanding how data and experience shape capabilities, and how capability feedback guides better data.
• Model-driven data construction. Exploring how models and agents can generate, refine, and organize learning material.
• Data infrastructure. Making data work more fine-grained, accessible, and responsive to feedback in the agentic era.

I am actively involved in data infrastructure development, primarily within the Data-Juicer ecosystem, which brings together (1) reusable operators and Elastic Juicer for efficient, fault-tolerant execution; (2) locally deployable Juicer for natural-language data refinement; (3) Data-Juicer Agents for workflow orchestration; and (4) Data-Juicer Sandbox / Panel for traceable experimentation and model feedback. Together, these components support my exploration of recursive self-improvement (RSI) in LLMs and agents. I also participate in research and explorations with Trinity-RFT and QwenPaw.

Feel free to get in touch for research discussions and collaboration.

Google Scholar / OpenReview

📣 News

• Aug 2026Our study on task learnability as a prior for LLM reinforcement learning is on arXiv.
• Jul 2026I joined Tongyi Lab, Alibaba Group, as a researcher.
• Jun 2026Our work on data refinement includes Juicer, a locally deployable model, and CDR-Bench, a benchmark for compositional recipe execution.
• Apr 2026GeoAlign was accepted to ICML 2026! Geometric rollout curation for robust LLM reinforcement learning.
• Mar 2026Attention Basin was accepted to ACL 2026! Understanding how contextual position affects LLMs.
• Jan 2026BOTS was accepted to ICLR 2026! Bayesian online task selection for reinforcement fine-tuning.
• Sep 2025DaaR was accepted to NeurIPS 2025! Diversity-driven data selection.
• Sep 2025MindGYM was accepted to NeurIPS 2025! Data synthesis for reasoning.

Conference papers

1. (ICML 2026) GeoAlign: Geometric Rollout Curation for Robust LLM Reinforcement Learning

   Ting Zhou, Zhenqing Ling, Yiyang Zhao, Ying Shen, Daoyuan Chen

2. (ACL 2026) Attention Basin: Why Contextual Position Matters in Large Language Models

   Zihao Yi, Zhenqing Ling^*, Delong Zeng^*, Haohao Luo, Zhe Xu, Wei Liu, Jian Luan, Wanxia Cao, Ying Shen

3. (ICLR 2026) BOTS: A Unified Framework for Bayesian Online Task Selection in LLM Reinforcement Finetuning

   Qianli Shen^*, Daoyuan Chen^*, Yilun Huang, Zhenqing Ling, Yaliang Li, Bolin Ding, Jingren Zhou

4. (NeurIPS 2025) Diversity as a Reward: Fine-Tuning LLMs on a Mixture of Domain-Undetermined Data

   Zhenqing Ling^*, Daoyuan Chen^*, Liuyi Yao, Qianli Shen, Yaliang Li, Ying Shen

5. (NeurIPS 2025) MindGYM: What Matters in Question Synthesis for Thinking-Centric Fine-Tuning?

   Zhe Xu^*, Daoyuan Chen^*, Zhenqing Ling^*, Yaliang Li, Ying Shen

6. (COLING 2025) Enhancing Factual Consistency in Text Summarization via Counterfactual Debiasing

   Zhenqing Ling^*, Yuexiang Xie^*, Chenhe Dong, Ying Shen

7. (KDD 2024) On the Convergence of Zeroth-Order Federated Tuning for Large Language Models

   Zhenqing Ling, Daoyuan Chen, Liuyi Yao, Yaliang Li, Ying Shen

Preprints

1. (arXiv 2026) Beyond Solvability: Task Learnability as a Static Prior for LLM RL Post-Training

   Ting Zhou^*, Zhenqing Ling^*, Daoyuan Chen, Qianli Shen, Yilun Huang, Ying Shen, Yaliang Li

2. (arXiv 2026) CDR-Bench: Evaluating Faithful Execution of Compositional, Order-Sensitive Data Refinement Recipes

   Yuchen Huang, Xiang Li, Zhenqing Ling, Sijia Li, Qianli Shen, Daoyuan Chen, Yi R. (May) Fung, Yaliang Li

🛠 Systems & open source

As a core contributor:

Data-Juicer

Working on reusable data operators, scalable execution, and data–model workflows for foundation models and agents.

As a participant in the official project teams, with work spanning code, data processing, evaluation, and feedback:

Trinity-RFT

Contributing to data workflows and evaluation for reinforcement fine-tuning, alongside research on online task selection.

QwenPaw

Participating in model-training research and experiments for personal AI assistants.

FederatedScope

Contributing FedMeZO, our work on memory-efficient zeroth-order federated tuning of large language models.