SDE

• Amazon.com SDE Intern
• Reduced compiling time from 20 minutes to 2 minutes by exploratory trial of multiple IDEs and generating makefile from online compiler.
• Improved the development process for the team by deploying a multi-thread debugging environment.
• Implemented a C socket system wrapper of C++, which greatly simply the initialization for TLS connection.
• Designed timer, thread, and network module in C abstraction, and integrated them with existing logging system.
• Pioneered the expansion of customer accessibility by allowing those operating with Mbed to utilize functions available through the AWS libraries to connect personal devices.

Software engineer Intern WeWork Jun. 2019-Aug. 2019
 Redesigned the workflow of CI/CD for migrating the service from Heroku to internal Kubernetes Cluster.
 Designed database migration plan for service migration to prevent downtime and data loss.
 Enabled the different sales teams to edit product detail by implementing the UPDATE API and integrated
authentication filter for access control in the Service based on Spring Boot.
 Enhanced internal API doc system to retrieve information dynamically from URL by refactoring the API sniffer using Ruby.

Software engineer Intern SAP Sept. 2017-Mar. 2018
 Optimized login process by implementing a Single-Sign-On mechanism based on Shibboleth for clients’ identity management system.
 Built a demo integrating Ethereum as persistent storage to clients’ identity management system.
 Reduced repeating operation in MS Project by implementing VBA scripts for different working scenarios.

Software engineer Intern CloudWiz Jul. 2017-Sept. 2017
 Implemented multiple metrics collectors scripts using Python for services such as RabbitMQ and PostgreSQL.
 Improved metrics collecting process through Nginx’s reverse proxy for client’s intranet systems.
 Designed the automatic upgrade mechanism for metrics collectors in clients’ distributed server using checksum of the latest collectors.

Software Engineer at Amazon Game Studios 08/2016 ~ 10/2017
Project: Game Services and Core Engine Technology
• Using C++ to construct new features with Lumberyard pipeline, allowing users to import files and folders without locally copying and pasting through directories. Using Qt5 framework to realize dialogs and menu options functionalities.
• Collaborated with product managers to design and implement the customer scoring system, including UI implementation with Qt5, dialog
prompting decision mechanism, and data streaming into backend data warehouse.
• Implemented a feature to create a unified UI and backend behavior for file drag and drop across all windows in our product, as well as an efficient and cross-platform directory visualization method.
• Applied the feature that flexibly collects and asynchronously sends customer telemetry data at various parts of the engine, including UI behaviors, hardware specification, and feature usages.
• Collaborated with QA members with owned written unit tests throughout the production period, verifying complex and unclear bugs for steps of the bugs’ reproductions and the solution confirmations, improving productivity of bug fixing time.
• Work with UX members to replace the old menu layout with the newly designed, implemented menu layout.

Health Informatics

• Psychological Regulation System Based on VR
• Designed an algorithm based on Yibin Wu’s brainwave and psychology research; completed a decision tree in analysis process, and achieved automatic selection of different videos according to users’ respective psychological conditions.
• Deployed a Java server, realized ability to store and retrieve data in self-built Alibaba Cloud database; presented users’ physiological data via curve graph in real time.
• Developed a method using Arduino to send data collected from brainwave and skin conductor sensors.

# Data Science

• Book Recommendation Website (Gelan Technology) Jun-Aug 2016
• Designed a website which can recommend books to users based on their previous favorites.
• Divided contents in the websites as slots and utilized Naïve Bayes algorithm to calculate the rating matrix for each website.
• Developed a scalable pipeline to process 203GB data using Hadoop and automated the batch processing pipeline
• Created the frontend of the website using HTML and CSS with Bootstrap Framework.

Data Scientist SoFi, CA
• Mined and analyzed data from company and third-party vendor databases to drive Homeloan digital marketing campaigns optimization by executing A/B testing and forging a lending product propensity model with Python, PostgreSQL and AWS, which amplified response rate by 12% on average.
• Established the customer lifecycle dashboard on Tableau to track all marketing touch points through the funnel and engineered a campaign measurement framework to boost upcoming campaigns success rates.
• Programmed SoFi’s first Homeloan awareness campaign by selecting markets, identifying the right channels, marketing budget optimization and potential customer profiling.
• Manipulated large dataset on AWS to perform analysis to optimize marketing strategies for growth and evaluate business hypothesis.
• Developed an uplifting model to measure net lift of Student Loan Refinance DM campaigns and leverage the result to drive upcoming campaigns action, which increased the response rate by around 20% on average.

Data Scientist

SAP, CA • Led in developing a recommendation system in production for Configure Price Quote product to optimize sale process with an ensemble of matrix factorization, sparse linear methods and evolution analysis etc., which achieved 3% revenue increase in cross sell and 2% in up sell on average.
• Provided data-driven strategies and business insights for SAP LeadRocket Marketing automation product to enhance the efficiency of email campaigns and sales events resulting in increased conversion rate of 5% in 3 months.
• Collaborated with product, and engineering teams to develop feedback system to monitor key metrics for SAP AI performance and published dashboards for SAP AI Products.

• Big Data Analytics on Wikipedia dataset with MapReduce (Gelan Technology) Jun-Aug 2016
• Utilized Terraform to create and manage Hadoop cluster (1 master node and 15 core node) on AWS EMR.
• Developed Data Filter and MapReduce Program in Java based on test-driven development with Junit and MRunit to process a large text dataset(128GB) from Wikipedia

• Ran MapReduce jobs on AWS EMR and analyzed output data on Jupyter Notebook with Pandas Library

• Machine Learning Engineer Internship
• Developed a neural machine translation system based on neural network architecture Transformer.
• Collected and preprocessed training data by Truecasing, Tokenization, and Normalization.
• Improved translation system performance by 4.5 BLEU score through Fine-tuning, Back-translation, Model Ensembling, and Reranking.
• Used Postman to do API testing, prepared analysis report by drawing comparisons between competing products such as Google Translate, Youdao Translate.
• Applied the system into commercial use, added an internal instant messaging app with live translation function.

# Product Manager

• Book Recommendation Website (Gelan Technology) Jun-Aug 2016
• Designed a website which can recommend books to users based on their previous favorites.
• Divided contents in the websites as slots and utilized Naïve Bayes algorithm to calculate the rating matrix for each website.
• Developed a scalable pipeline to process 203GB data using Hadoop and automated the batch processing pipeline
• Created the frontend of the website using HTML and CSS with Bootstrap Framework.

# Quantative Analysis

• Book Recommendation Website (Gelan Technology) Jun-Aug 2016
• Designed a website which can recommend books to users based on their previous favorites.
• Divided contents in the websites as slots and utilized Naïve Bayes algorithm to calculate the rating matrix for each website.
• Developed a scalable pipeline to process 203GB data using Hadoop and automated the batch processing pipeline
• Created the frontend of the website using HTML and CSS with Bootstrap Framework.

Quantitative Research Analyst Trimtabs Investment Research, CA • Created the new investment strategies by analyzing the impact of large amount new offerings and new cash takeovers on S&P 500 performance and achieved 25% uplift for profit in a 5 year window. • Developed five model portfolios (U.S equity ETF model, NASDAQ model, country model etc.) to provide quantitative analysis about the stock market, which also beat the benchmarks by 2% on average. • Conducted researches on the different stock sectors performance to provided data-driven decisions for investment.

• Quantitative Analyst (Gelan Technology) Jun-Aug 2016
  • Developed a faster method for big data’s rolling polynomial regression by applying the expansion of normal equations, increasing the effeciency by over 200 times than the original method.
  • Refactored a function to fill in missing financial data in comapny’s database, increase effeciency by turning loop operations into column operations and improved filling accuracy by seasonal decomposition.
  • Built a model based on AR-OLS to predict the dispersion of stock market according to its autocorrelation and its relationship with stock markets’ volatility, the average residual is less than half of the traditional statistical models.
  • Quantitatively defined 47 different form factors and back-tested them to calculate excess returns and classified them according to their performance.

  Skills

  Python: pandas, seaborn, matplotlib, scikit-learn, dask, sqlalchemy, sqlite3
  R: future, ggplot, RSQLite, R6, devtools, stringr, dplyr SQL: skilled in writing SQL query to filter, sort, and summarize data Others: TABLEAU, AB Testing, SPARK
  Language: Java, JavaScript, TypeScript, Ruby, Python, PHP, C/C++, HTML/CSS, SQL, Shell, MATLAB, VBA Frameworks & Tools: Spring Boot, Docker, Kubernetes, Ruby on Rails, React, Angular, Node.JS, Django, Spark