TL;DR A bounded-exploration, verification-gated framework for improving complete agentic systems, instantiated in XYZ-Aquila, a family of open-weight Deep Search agents. XYZ-Aquila-pro records the highest reported score among all listed systems on BrowseComp-ZH, LiveBrowseComp, and WideSearch.
Hi, I am Yifei Zhang (张逸飞), a master's student in Financial Engineering and Fintech at Nanjing University's School of Management & Engineering, advised by Prof. Honghai Yu. I hold a bachelor's degree in Computer Science and Technology from the Department of Computer Science and Technology at Nanjing University, where I specialized in Computer & Financial Engineering Experimental Class (a Top Talent Program).
I am currently a research intern at Century Frontier, working on AutoResearch in the quantitative area. Before that, I was a research intern at XYZ AI LAB from April 2026 to July 2026, advised by Xiao Yang and Jiang Bian, focusing on AI4AI research on LLM Post-training. Before that, I was a research intern at Microsoft Research Asia (MSRA) from May 2025 to January 2026, advised by Xiao Yang, focusing on Autonomous Agent research. Previously, I had the privilege of collaborating with Prof. Benyou Wang and Prof. Qianqian Xie on exploring Large Language Models in Finance.
I have published at top-tier conferences including NeurIPS, ICML, and ACL. Notably, one of my papers received the Best Paper Award at the ICLR 2025 Workshop on Advances in Financial AI, where I presented it as an oral presentation in Singapore. My research interests are AI4AI and Autonomous Agents.
🎯 Open to opportunities: I am currently on the job market for full-time positions. If you think I would be a good fit, feel free to reach out!
📄 More about me:
English Resume
中文简历
01
News
- 07/2026 🚀 Open-sourced XYZ-Aquila, a family of Deep Search agents, together with the agent runtime harness AxisAgentic.
- 05/2026 🏆 Awarded the Kweichow Moutai "Pillars of the Nation" Talent Program (only 5 university-wide).
- 05/2026 📄 One paper accepted by ICML 2026. You can find the paper here.
- 04/2026 📄 One paper accepted by ACL Findings 2026. You can find the paper here.
- 10/2025 🏆 Receive the National Scholarship at Nanjing University.
- 09/2025 📄 One paper accepted by NeurIPS 2025. You can find the paper here.
- 08/2025 🎖️ One paper selected as the Outstanding Paper Award for the 7th Conference of the Chinese Society of Optimization, Overall Planning, and Economic Mathematics.
- 05/2025 🚀 Start my internship at MSRA-Beijing!
- 04/2025 🏅 Give a talk at Financial AI @ ICLR 2025 and won the Best Paper Award (1 out of 53). You can find the paper here.
- 01/2025 📝 One paper accepted by NAACL Findings 2025. You can find the paper here.
- 09/2024 🎓 Begin new journey as a master student at Nanjing University.
- 07/2024 🔬 Start a RA position at CUHK(SZ) with Prof. Benyou Wang.
- 06/2024 🎉 Graduate from Nanjing University!
1 / 3
RD-Agent Project
A unified research line on autonomous ML engineering agents. The technical report defines the RD-Agent framework, while the two follow-up papers study how the same project scales to Machine Learning Engineering (MLE) optimization and autonomous LLM fine-tuning.
R&D-Agent: An LLM-Agent Framework Towards Autonomous Data Science
(Technical Report)
A top-performing open-source framework on MLE-Bench that structures autonomous data science into two parts: Research for proposing ideas and Development for turning them into executable solutions.
Reasoning as Gradient: Scaling MLE Agents Beyond Tree Search
(MLE Setting of RD-Agent)
The MLE setting of RD-Agent, where Gome sets the strongest open-source MLE result on MLE-Bench under a closed-world protocol. Its main finding is that as model reasoning capabilities improve, gradient-based optimization increasingly outperforms gradient-free tree search.
FT-Dojo: Towards Autonomous LLM Fine-Tuning with Language Agents
(Fine-Tuning Setting of RD-Agent)
The fine-tuning setting of RD-Agent, introducing an interactive benchmark with 13 tasks across 5 domains and FT-Agent, which achieves the best results on 10 of 13 tasks.
TL;DR A multi-agent simulation platform for financial markets that integrates behavioral economics and social network dynamics, enabling large-scale market experiments with diverse agent behaviors.
03
Education
Nanjing University
2024.09 – Present
Master's Student in Financial Engineering and Fintech
School of Management & Engineering | Advised by Prof. Honghai Yu
Nanjing University
2020.09 – 2024.06
Bachelor in Computer Science and Technology
Computer & Financial Engineering Experimental Class (Top Talent Program)
04
Internships
Century Frontier
2026.07 – Present
Research Intern
Working on AutoResearch in the quantitative area.
XYZ AI LAB
2026.04 – 2026.07
Research Intern
Focusing on AI4AI research on LLM Post-training | Advised by Xiao Yang and Jiang Bian
Microsoft Research Asia (MSRA)
2025.05 – 2026.01
Research Intern
Focusing on Autonomous Agent research | Advised by Xiao Yang
CUHK-Shenzhen NLP Group
2024.07 – 2025.05
Research Assistant
Exploring Large Language Models in Finance | Advised by Prof. Benyou Wang
