关于我

我是一个多模态大模型和AI智能体研究者,作为核心贡献人员,已经完成1篇论文、3个开源项目代码、1套Qwen微调模型以及1套数据集。我与我的其他研究伙伴共同创立了GolGrin Research,这是一个针对人工智能、金融科技开展产业研究的初创型研究组织,目前我们正在开展关于大模型推理上市公司衍生品策略与信用风险、企业延迟付款、库存预测等方面的研究,文章成果也将投稿到国际顶级期刊或者会议。过去,我在香港科技大学(广州)AgenticFin Lab参与研究工作,受陈思佳教授指导,研究方向为多模态大模型推理、因果推断和资产定价,我们的文章即将投稿到ACL、EMNLP等国际顶会。我曾经也在深圳大学受尼娜教授、吴未教授、陈鑫教授、石洋教授等导师的学术指导,在人工智能、金融科技、国际交流等领域进行学术协助与研究。

代表性论文
Yuqun Zhang, Yuxuan Zhao, Sijia Chen arXiv
一个前沿框架,通过对抗性智能体和MCTS机制增强视觉语言模型对复杂金融图像的理解,解决金融图像分析中的独特挑战,包括数据稀缺、领域复杂性和层次推理需求。
开源项目
PyFi框架代码、数据集与微调模型 · 核心贡献者
  • 特点:金字塔式架构、对抗性智能体、大规模数据集、综合评估基准、模块化设计、针对金融图像理解微调的Qwen模型
  • 第一作者
FinMycelium · 核心贡献者
  • 一个金融事件重构平台项目,能够从多源、异构的公开文档中,将完整的金融事件过程重构为结构化的时间线。该平台基于大模型驱动的多智能体系统构建,各智能体协同工作,完成对大规模、异构、含噪声的真实世界数据的收集、匹配与摘要,最终构建出全面且结构化的金融事件重构结果。
CIJUNTalker · 创始人兼核心贡献者
  • 一个以多模态大模型为核心的开源项目,编排多个智能体(导演、图像生成、语音合成、剪辑、PPT解析、脚本生成等),实现全自动视频制作:从复杂的图像/视频理解到静态/动态视频生成、PPT转视频、Vlog制作乃至电影级长视频生成。
荣誉 & 奖励
拔尖创新人才奖一等奖(前20%,两次)
微众奖学金一等奖(前2%)
荔园之星(前2%)
深圳大学优秀实习生(前4%)
腾讯“锐意挑战创新团队”奖
“双创之星”一等奖(前2%)
优秀学生干部一等奖(前2%,两次)
学院优秀本科毕业生(6/57)
研究 & 实习
香港科技大学(广州) · 研究助理、博士生候选人 (2025.07 - 至今)
  • 导师陈思佳教授,加拿大多伦多大学电子与计算机工程系博士,主要研究方向为分布式计算、多模态学习、决策算法、大模型等
  • 我在金融科技学域AgenticFin Lab参加研究工作,受陈思佳教授指导,研究方向为多模态大模型推理、因果推断与资产定价。
微众银行 · 创新研究岗实习生 (2023.02 – 2023.04)
  • 在科技创新产品部创新研究与标准化室参加实习,参与金融科技、数字银行、元宇宙相关研究工作。
深圳大学(吴未教授) · 研究与技术助理 (2022.10 – 至今)
  • 导师吴未教授,英国伯明翰大学博士,牛津大学助理研究员,深圳大学助理教授,主要研究方向为金融科技、全球数字经济治理、创新城市网络等。
  • 参与国际交流、人工智能研究项目,提供人工智能技术支持,撰写代码为项目开展提供必要基础工具。
深圳市金融稳定发展研究院 · 研究实习生 (2022.07 - 2022.10)
  • 在人才培训部参加实习,参与深圳市金融领军人才培训相关研究和落地工作。
创新创业
金绿数智GolGrin GolGrin Research · 联合创始人 (2023.08 – 至今)
  • 一家人工智能初创公司,为各类企业、事业单位、政府部门、公益组织提供AI模型、智能体与软件技术服务并开展研究工作。
  • 例1:开发大模型工作流服务用于公益组织的工作流程和项目,助力工作人员在工作考核中通过率达90%以上
  • 例2:开发、训练手写字OCR模型并将配套软件部署在国产系统与海光GPU上,经验收单字识别率高于99%,白板手写识别率高于90%,任意文件识别响应速度小于100毫秒
此君智能 CIJUN AI · 创始人(2026.01 - 至今)
  • 开发训练面向机器人服务领域的多模态大模型与智能体,服务体系涵盖交易商城、社区、论坛、知识库等组成部分
报告交流
AI重构社会力量:从“工具革命”到“生态革命”
  • 受邀报告:系统阐述AI如何从工具应用演进为重塑社会组织服务生态的关键力量,并展示了其在重构服务模式、拓展服务边界方面的多元技术路径。
人工智能如何改变我们的日常
  • 受邀报告:系统拆解大语言模型原理,并围绕大模型个性化应用、智能体、AI设计生成等十余项典型技术实践,深度解析了AI在办公、创作与生活场景中的高效落地路径。
人工智能前沿应用原理及实用工具简析
  • 受邀报告:系统梳理ChatGPT的发展脉络与底层逻辑,深度解析了国内外主流大语言模型工具的使用方法,并结合人工智能法律法规与产业融合趋势,为观众提供了从技术原理到高效应用落地的完整方法论。
赴港在大湾区青年论坛发言并作主题报告
  • 受邀报告:分享关于创新创业、人工智能、ESG等领域的见解
技术能力
深度学习框架
  • Github 协作框架 / HuggingFace Transformers
  • PyTorch / DeepSpeed / Megatron-LM / PEFT
编程语言 & 工具
  • Python / C++ / CUDA
  • LaTeX / Markdown
About Me

I am a researcher specializing in multimodal large language models and AI agents. As a key contributor, I have completed 1 paper, 3 open-source project codebases, 1 Qwen fine-tuned model, and 1 dataset. Together with some of my research partners, I co-founded GolGrin Research, a startup research organization focused on industrial research in artificial intelligence and financial technology. We are currently conducting research on areas such as listed companies’ derivative strategies and credit risk based on large model reasoning, corporate payment delays, and inventory forecasting, with the aim of submitting our findings to top-tier international journals or conferences. Previously, I participated in research at the AgenticFin Lab at The Hong Kong University of Science and Technology (Guangzhou), advised by Professor Sijia Chen, focusing on multimodal large model reasoning, causal inference, and asset pricing. Our paper is planned for submission to top-tier international conferences such as ACL and EMNLP. I have also previously received academic guidance and engaged in academic assistance and research support in artificial intelligence, financial technology, and international exchange advised by Professor Nina, Wei Wu, Xin Chen, Yang Shi, and other mentors at Shenzhen University.

Selected Publications
Yuqun Zhang, Yuxuan Zhao, Sijia Chen arXiv
A framework to enhance VLMs in understanding complex financial images via adversarial agents and MCTS, addressing data scarcity, domain complexity, and hierarchical reasoning.
Open Source Projects
PyFi Framework Code, Dataset & Fine-tuning Models· Core Contributor
  • Features: pyramid-like architecture, adversarial agents, large-scale dataset, comprehensive benchmark, modular design, fine-tuned models for image understanding.
FinMycelium · Core Contributor
  • A Financial Event Reconstruction Platform that reconstructs the complete financial event process as a structured timeline from multi-source, diverse public documents. It is built on a large model–based multi-agent system, in which agents cooperate to collect, match, and summarize large-scale, heterogeneous, and noisy real-world data, ultimately building a comprehensive and structured reconstruction of the event.
CIJUNTalker · Founder & Core Contributor
  • An open-source project for automated video production using multi-modal LLMs and multiple intelligent agents (director, image generation, speech synthesis, editing, PPT parsing, script generation), enabling image/video understanding, static/dynamic video generation, PPT-to-video, Vlog, and movie-grade long video generation.
Honors & Awards
Top Innovative Talent Award First Prize (top 20%, twice)
WeBank Scholarship First Prize (top 2%)
Liyuan (Shenzhen University) Star (top 2%)
Shenzhen University Outstanding Intern (top 4%)
Tencent "Determined Innovation Team" Award
Outstanding Student Leader First Prize (top 2%, twice)
Innovation Star First Prize (top 2%)
College Outstanding Undergraduate Graduate (6/57)
Research & Internship
The Hong Kong University of Science and Technology (Guangzhou) (2025.07 – present)
  • Research Assistant, PhD Student Candidate
  • Professor Sijia Chen, my mentor, holds a Ph.D. from the Department of Electrical & Computer Engineering at the University of Toronto. His main research interests include distributed computing, multimodal learning, decision-making algorithms, and large language models.
  • I participated in research at AgenticFin Lab in the Financial Technology Thrust, under the supervision of Professor Sijia Chen. My research focuses on multimodal large language model reasoning, causal inference, and asset pricing.
WeBank (2023.02 – 2023.04)
  • Innovation Research Post
  • Internship in the Innovation Research and Standardization Office of the Technology Innovation Products Department, working on fintech, digital banking, and metaverse research.
Shenzhen University (Advised by Prof. Wei Wu) (2022.10 – present)
  • Research&Technical Assistant
  • Professor Wei Wu, my mentor, holds a Ph.D. from the University of Birmingham, UK. He serves as an Assistant Professor at Shenzhen University and a Research Associate at the University of Oxford. His main research interests include financial technology, global digital economy governance, and innovative urban networks.
  • I participated in international exchange and AI research projects, provided AI technical support, and wrote code to develop essential infrastructure tools for project implementation.
Shenzhen Financial Stability & Development Institute (2022.07 – 2022.10)
  • Project Research Post
  • Internship in the Talent Training Department, participating in research and implementation of Shenzhen financial leadership talent training programs.
Innovation & Entrepreneurship
GolGrinGolGrin Research · Co-founder (2023.08 - present)
  • An AI startup providing AI models and agent technical services to enterprises, institutions, government agencies, and NGOs.
  • Case 1: Developed LLM workflow and software services for NGO workflows, helping staff achieve a 90%+ pass rate in performance reviews.
  • Case 2: Developed and trained a handwriting OCR model with software deployed on domestic systems and Hygon GPUs, achieving >99% single-character recognition, >95% whiteboard handwriting recognition, and <100ms response time for any document.
CIJUN AI · Founder (2026.01 - present)
  • Develop and train multimodal large language models and intelligent agents for the field of robot services, with a service system covering components such as transaction malls, communities, forums, and knowledge bases
Technical Talks
AI Reconstructing Social Power: From "Tool Revolution" to "Ecosystem Revolution" (2025)
  • Invited talk: Explained how AI evolved from a mere tool application into a key force reshaping the social organization service ecosystem, and demonstrated the diverse technological pathways for reconstructing service models and expanding service boundaries
How AI Changes Our Daily Life (2024)
  • Invited talk: Deconstructed the principles of large language models, and, focusing on over a dozen typical technical practices such as personalized LLM applications, intelligent agents, and AI-powered design generation, deeply analyzed the efficient implementation pathways of AI in office, creative, and daily life scenarios
Principles of Cutting-Edge AI Applications and Practical Tools (2023)
  • Invited talk: Outlined the development trajectory and underlying logic of ChatGPT, deeply analyzed the usage methods of mainstream large language model tools both domestically and internationally, and, combined with AI-related laws, regulations, and industry integration trends, provided the audience with a complete methodology ranging from technical principles to efficient application implementation
Speech and keynote report at the Greater Bay Area Youth Forum in Hong Kong (2023)
  • Invited talk: Share insights on innovation and entrepreneurship, artificial intelligence, ESG, and other fields
Technical Skills
Frameworks
  • Github Collaboration Framework / HuggingFace Transformers
  • PyTorch / DeepSpeed / Megatron-LM / PEFT
Languages & Tools
  • Python / C++ / CUDA
  • LaTeX / Markdown
欢迎探讨科研合作,随时邮件联系。 Open for research collaboration