|
I am a 3rd year PhD candidate in Statistics at the University of Texas at Dallas (UTD).
I used to work as a full time Researcher/Data Analyst at Fortune 500 Companies (Bank of China, Taikang, China Merchants, etc.).
My research interests are Bayesian Modeling, Bioinformatics, and Machine Learning
I passed CFA EXAM Level III - Charter pending.
Email  / 
GitHub  / 
LinkedIn
|
|
Work Experience
Bloomberg, New York, NY
Quantitative Machine Learning Intern • June 2024 to Present
- Accelerate the derivative pricing process with deep neural networks (DNN) in PyTorch
|
|
UT Dallas, Dallas, TX
Research Assistant • Jan 2023 to Present
Advisor: Qiwei Li
|
|
Bank of China, Beijing, China
Industry Researcher • July 2019 to Aug 2020
- Analyzed bank customer service data by the NLP Semantic analysis system, extracted effective information by keyword
- Detected and predicted frequently happened customer issues, and generated over 30 reports
|
|
Beijing Micai Investment Co., Ltd., Beijing, China
Bitcoin and Crypto Data Analyst • July 2018 to Mar 2019
- Implemented automatic quantitative trend-tracing trading strategy with Support Vector Regression(SVR)
- Applied web-crawler with Python to collect online STO data then saved to the database by MYSQL
- Generated bilingual research reports on the in-depth study of new-launched STO cryptocurrencies
|
|
Taikang Pension & Insurance Co., Ltd. , Beijing, China
Analysis Intern • Mar to May 2018
- Made strategic asset allocation decisions with B-L model in MATLAB with data from 2013 to 2018
- Supervised the investment performance of investment of subordinate bodies and gave improvement advice
|
|
China Merchants Securities Co., Ltd., Shenzhen, China
Quantitative Analysis Intern • June to Aug 2017
- Designed investment strategies by applying LSTM to historical data, with indicators such as RSI and MACD
- Generated reports on the study of newly listed companies and companies currently under procedure to be listed
- Detected listed companies with default risk with financial-soundness indicators in XGBoost and reached an 80%
backtesting accuracy
|
|
Projects
BayeSMART: Bayesian clustering of multi-sample spatially resolved transcriptomics data
- Preprocessing and Analysing tissue images by deep neural networks (DNN) models in PyTorch
- Achieving an accuracy of 90% in detecting cancer regions and improving up 54% over the state-of-the-art methods
- Developing parallel computing and sampling algorithms for GPU clusters
GitHub: BayeSMART
|
|
Distantly-Supervised Joint Entity and Relation Extraction with Noise-Robust Learning
- Incorporated a pre-trained transformer into sequence tagging scheme for distantly-supervised joint extraction
- Proposed a bootstrap learning framework with a noise-robust loss to select high-quality instances dynamically
Paper: DENRL
|
|
Bayesian Dirichlet-Multinomial Model for Integration of Clinical and Single Cell Data
- Proposed a hierarchical Bayesian framework for the integrative analysis of cell-type abundance count data and covariates analysis
- Implemented Markov Chain Monte Carlo(MCMC) methods with R and C++
- Obtained the discriminating cell types that are aligned with the findings in the original papers, with new discoveries of the relationship between the abundance of cells and covariates at the cell-type level that were not studied before
GitHub: Bayesian-DM-Regression
|
|
Change-point Detection Using Bayesian Inference
- Applied t-shrinkage prior and Horseshoe prior to historical DJI return and Bitcoin-USD return during fluctuating time
- Implemented Markov Chain Monte Carlo(MCMC) methods with R and visualized the detected change-points
- Detected 100% change-points successfully by the given criteria to define a change-point
GitHub: Change-point Detection for Financial Markets
|
|
Potential ETC Customer Identification
- Processed and cleaned the original customer dataset of 9 million samples with de-noise analysis
- Applied naive Bayes principle to obtain the soft voting output under the assumption of independence
- Utilized data discretization methods to further reduce model complexity
- Refined the data by Grid Search and Ensemble Generation and achieved an AUC by 0.9 of targeting a potential ETC Customer (rank 10/200)
|
|
Feature Sensitive 3D Printing Adaptive Slicing Algorithm
- Built the feature sensitive metric of the object surface, then mapped the 3-dimensional surface points to a sextuple space
- Pinpointed areas with significant normal vector change and huge curvature of local surface
- Traversed all the layers and pairs of points on the layer, which greatly improved the surface accuracy by 10%
|
|
|
Teaching Experience
Math 2419 Calculus II, University of Texas at Dallas
In person lecture: 4 hours per week
|
Aug - Dec 2022
|
Math 2417 Calculus I, University of Texas at Dallas
In person lecture: 4 hours per week
|
Aug - Dec 2021
|
CSA Calculus Tutor, Columbia University
|
Sep - Dec 2017
|
|
Honors & Awards
- Best Poster Award, The Second UT Dallas/UT Southwestern Annual Workshop on Imaging and Data Science, Richardson, Texas
- Travel Award, 2023 Southern Regional Council on Statistics Summer Research Conference, Waco, Texas
- Nomination Award in 2019 Bank of China Machine Learning Modeling Contest, Beijing, China
- Outstanding Work Prize in 2016 Student Entrepreneurship and Innovation Competition, Shanghai, China
- ECUST academic scholarship for 2012-2016 consecutively during undergraduate, Shanghai, China
|
|
| |