Bowen Zhang | 张博文 B.S. in Data Science, CUFE

Central University of Finance and Economics

Bowen Zhang 张博文

Data Science undergraduate at CUFE, exploring fintech through financial data, statistical modeling, and machine learning.

Research [email protected] Blog
Portrait of Bowen Zhang

About

Exploring FinTech with Data Science

I am a B.S. student in Data Science and Big Data Technology at the School of Information, Central University of Finance and Economics. I ranked 1st in my cohort (1/21) and am currently exploring how statistical learning, econometrics, and data systems can be used in fintech.

  • Financial data modeling: prediction, ranking, and risk-oriented modeling with Python, SQL, and tree-based machine learning.
  • Experiment and policy evaluation: causal inference, randomized experiments, IV estimation, and variance-aware empirical design.
  • Data infrastructure: reproducible workflows, large-scale Spark analysis, and research tooling for applied analytics.

Research

Research Foundations

Working Paper First Author

分层两阶段随机试验下的因果推断

Advisor: Yuehan Yang

Builds a stratified Complier Average Direct Effect estimator for two-stage randomized experiments, with direct relevance to empirical evaluation and experiment design.

  • Proved unbiasedness of the stratified ITT estimator and consistency with asymptotic unbiasedness of the CADE estimator.
  • Established equivalence between the stratified estimator and weighted two-stage least squares under specific weights.
  • Derived conservative variance estimators and implemented R simulations for balanced and unbalanced cells.
In Submission Third Author / Co-Contribution

基于基础模型与多模态药物编码的单细胞扰动预测

Manuscript currently in submission to a Nature sub-journal

A machine learning research project outside finance that strengthened my experience with sparse high-dimensional prediction, multimodal representation learning, and robust evaluation.

  • Designed multimodal drug fusion with RDKit descriptors, GIN molecular graphs, Swin structure images, and two-stage cross-attention.
  • Built sparse-signal-aware losses with control-normalized adaptive weighting, direction consistency, and MMD distribution alignment.
  • Implemented scGen baseline replication and per-perturbation evaluation covering Pearson, Pearson-Delta, Pearson-DEG, R2, MSE, and Sinkhorn metrics.

Selected Projects

Applied Modeling and Data Systems

Kaggle Bronze Top 8%

水平井地层厚度曲线预测

Built leakage-aware grouped cross-validation, compared Ridge, LightGBM, and CatBoost, and improved public leaderboard RMSE from 8.860 to 8.055 through residual modeling and ensemble learning.

Spark 55M Records

PySpark 电商用户行为大规模分析与购买倾向预测

Used Hive, Spark SQL, Sqoop, and MLlib to analyze 55 million behavior logs, build warehouse-style aggregations, and train a distributed purchase-intent baseline with AUC 0.71.

Experience

Internships and Research Assistance

Fit2Cloud · Marketing Intern

2025.8-2025.12 | Beijing

Tracked DataEase GitHub Issues and community feedback, summarized high-frequency user needs, and reported product insights to the team.

Northeast Securities · Strategic Research Intern, Investment Banking

2025.7-2025.8 | Beijing

Researched private targets in underwater acoustic signal processing and marine sensing, using Wind to evaluate financial structure and growth potential.

MIIT Big Data Industry Talent Base (Zhejiang) · Research Assistant

2024.11-2024.12 | Remote

Supported data and industry research tasks in a remote research-assistant role.

Skills

FinTech-Relevant Toolkit

Programming

Python, SQL, R; Java, C, C++; Hadoop, Spark, Hive, Kafka, Sqoop.

Empirical Methods

RCT, IV, PSM, DID, RD, uplift modeling, AB-test design, econometrics, and causal inference.

Machine Learning

Feature engineering, model validation, tree ensembles, PyTorch, representation learning, and reproducible simulation pipelines.

Contact

Bowen Zhang | 张博文

School of Information, Central University of Finance and Economics, Beijing, China.