Siyuan (Sven) LI
PhD StudentMy research focuses on building reliable LLM systems for high-stakes legal and financial applications. I am particularly interested in structured representations, agentic workflows, and regulatory reasoning, together with evaluation methods that detect omissions, factual errors, and uncertainty in complex model outputs.
HKUST(GZ) · Guangzhou, China
Overview
Research
My research centers on LLM systems for high-stakes legal and financial settings, with a focus on agents, knowledge graphs, compliance, report generation, and automated financial research.
Selected publications4
Siyuan Li*Jian Chen*Rui YaoXuming HuPeilin ZhouWeihua QiuSimin ZhangChucheng DongZhiyao LiQipeng XieZixuan Yuan
The first large-scale Chinese dataset for financial regulatory compliance with an automated checking pipeline.
KDD 2026 Dataset & Benchmark Track, Poster
Jian Chen*Siyuan Li*Chucheng WanZixuan Yuan
A span-grounded legal parsing framework for rules, exceptions, and scope relations.
KDD 2026 Dataset & Benchmark Track, Poster
KnowMT-Bench: Benchmarking Knowledge-Intensive Long-Form Question Answering in Multi-Turn Dialogues
2025Junhao Chen*Yu Huang*Siyuan Li*Rui YaoHanqian LiHanyu ZhangJungang LiJian ChenBowen WangXuming Hu
A benchmark for multi-turn long-form question answering in knowledge-intensive domains.
arXiv preprint
Rui YaoQi ChaiJinhai YaoSiyuan LiJunhao ChenQi ZhangHao Wang
An uncertainty-aware framework for interpreting Federal Reserve communications with more reliable policy-stance predictions.
AAAI 2026, Oral
Projects1
A four-skill open-source toolkit for PowerPoint, Word, web-demo video, and social-media content generation.
Open SourceAgent SkillsPowerPointWordWeb Demo VideoSocial Media
Work experience
Lingyue Technology Co., Ltd. (AI4Finance Startup)
AI Algorithm Engineer
Developed and delivered a compliance component for an investment-advisory system, organized the existing compliance framework into a hierarchical rule structure, and built data pipelines for heterogeneous financial data, company knowledge graphs, and downstream quantitative research.
Credit Bond Research Assistant
Led the development of a real-time credit risk assessment system with Python and SQL, used diversified financial data and interpretable models such as XGBoost to predict credit-bond default risk, and replicated fixed-income studies from top journals.
Industry Research Institute Research Assistant
Contributed to institutional and internal research reports on elderly-care services, built a cohort-component forecasting model for population size and age structure, and analyzed around 1,000 products from the Shanghai Senior Care Expo.
Education
PhD in Artificial Intelligence
LLM applications in law and finance, especially agents, evaluation, and quantitative trading.
MPhil in Artificial Intelligence
LLM applications in law and finance, especially compliance, structured reasoning, and report generation.
BA in Finance
GPA 3.6/4.0, with coursework in microeconomics, macroeconomics, financial accounting, intermediate financial accounting, econometrics, and corporate finance.