University of Washington - B.S. in Mathematics

Building practical ML systems for quantitative finance.

I am Weikun Zhang, a math student focused on quant research, machine learning, and applied AI engineering.

  • Quantitative research and forecasting pipelines
  • NLP and RAG systems for finance workflows
  • RL for arithmetic circuits and symbolic optimization
See Published Work Work With Me

About

Quick background

Education

University of Washington | Sep 2022 - Jun 2026

Bachelor of Science in Mathematics

Focus: quantitative modeling, machine learning, and optimization.

Core courses: Stochastic Process; Stochastic Calculus and Option Pricing; Numerical Analysis; Markov Chain; Linear Optimization; plus ML/AI, software engineering, real analysis, and databases.

What I bring

  • Strong mathematics with practical engineering delivery
  • Hands-on ML modeling, validation, and backtesting
  • Reliable Python + SQL automation for production workflows

Experience

Internships

Sep 2025 - Present | Shanghai, China

Quantitative Research Intern, Aegon-Industrial Fund Management

  • Built an end-to-end ROE forecasting pipeline: data, training, validation, and backtesting.
  • Improved out-of-sample performance with cross-validation and regularization.
  • Automated daily reports for research and portfolio teams using Python and SQL.

Jun 2024 - Oct 2024 | Shanghai, China

Software Engineer Intern, CIB Financial Technology Department

  • Built NLP + RAG tagging pipelines for investment research workflows.
  • Improved compliance response rate by 16% through model rule debugging.
  • Raised research information processing efficiency by 30%.

Jul 2023 - Oct 2023 | Shanghai, China

Software Engineer Intern, Tencent IEG

  • Created reusable Python scripts to reduce repetitive QA work.
  • Prepared JSON/CSV test data with SQL and regex support.
  • Executed and tracked 20+ functional tests in Jira.

Projects

Selected research projects

RL for Arithmetic Circuits / Polynomial Factorization

Core Member, UW Math AI Lab | Dec 2024 - Present

Built an RL + GNN + MCTS framework with SymPy and PyTorch Geometric; reduced expression complexity by 10%-30%.

Adversarial Attacks on Multi-Armed Bandits

Individual Student Researcher | Jul 2024 - Apr 2025

Built a simulation platform to benchmark ETC, UCB, and Thompson Sampling in adversarial, non-stationary settings.

Jane Street Quant Trading Competition

Sep 2024 - Dec 2024

Engineered alpha features with LightGBM/XGBoost and strict time-series validation; ranked top 10% worldwide.

Achievements

Selected highlights

Jane Street Quant Trading Competition

2024

Ranked in the top 10% globally among more than 10,000 participating teams.

Measurable Internship Impact

CIB Financial Technology Department

Improved compliance response rate by 16% and boosted information processing efficiency by 30%.

Published Work

CircuitBuilder: From Polynomials to Circuits via Reinforcement Learning

arXiv:2603.17075 | Submitted Mar 2026

We frame arithmetic circuit discovery as a single-player RL problem and compare PPO+MCTS with SAC. SAC performs strongly on simpler settings, while PPO+MCTS scales better to harder cases.

Co-authored with collaborators from UW Math AI Lab and partner researchers.

View arXiv Read PDF
Efficient and inefficient circuit structures
Efficient vs. inefficient circuit structures from the arithmetic circuit research workflow.
Circuit figure will appear here once the image is available.

Skills

Technical toolkit

Python Java JavaScript SQL MATLAB PyTorch Geometric LightGBM XGBoost Git Tableau Excel (VBA / Power Query) React Native

Contact

Let's connect