I am a UC Berkeley Data Science student interested in applied AI, machine learning, and software development.
My experience spans AI-assisted research, accessible software, and financial data analysis. I am seeking opportunities in AI and data-related fields. Contact me at jingxuan.lyu@berkeley.edu.
MiraclePlus | Beijing, China
Built an AI-assisted workflow with Python, n8n, and LLM APIs to organize 1,500 candidate profiles across 12 industries. Designed contact deduplication, candidate scoring, human approval, CRM tracking, and 300+ personalized outreach drafts with review checkpoints.
Chinese Academy of Sciences | Beijing, China
Designed an accessibility-focused calculator with streamlined input logic for students with visual impairments, then iterated on the interface through user testing with students at Beijing School for the Blind.
London Stock Exchange Group | Beijing, China
Analyzed customer and transaction datasets with Python and Azure SQL for credit-risk and customer-risk workflows, and built reusable data-cleaning and transformation pipelines for downstream analysis.
B.A. in Data Science
Relevant coursework: Data Structures (CS 61B), Discrete Mathematics and Probability Theory (CS 70), Structure and Interpretation of Computer Programs (CS 61A), and Foundations of Data Science (Data C8).
Research items remain separate from projects and keep their own pages.
Machine Learning Research
A fraud-detection study comparing classical ML models, ensemble methods, and a CNN on large-scale transaction data.
Machine Learning Research
A research project on machine learning models for credit risk prediction, developed through advanced study in algorithms and data.
Time-Series Forecasting Research
A weather-forecasting study using high-frequency campus sensor data to support short-term rainfall prediction.
Selected work across applied AI, data, and software systems.
AI & Financial Data
Prototyped an LLM-based assistant for analyzing 120 days of transaction history, including recurring charges, spending categories, and cash-flow patterns. Its tool-calling architecture retrieves only the relevant transaction subsets on demand.
Web Engineering
Built and deployed this responsive portfolio with reusable components, custom routing, a custom domain, and Git-based version control.
Machine Learning
Built and compared logistic regression, random forest, and CNN models for imbalanced fraud data, using feature engineering and precision, recall, F1 score, and confusion matrices to evaluate robustness.
Data Structures (CS 61B) · Discrete Mathematics and Probability Theory (CS 70) · Structure and Interpretation of Computer Programs (CS 61A) · Foundations of Data Science (Data C8)