Yun-sen Gu

Los Angeles, CA *****, Telephone: 213-***-****, Email: adb0b2@r.postjobfree.com, LinkedIn: linkedin.com/in/ysgu

Summary: First-year graduate student specializing in data analysis and machine learning. Has solid mathematics background and 5 years of experience in programming.

EDUCATION

Master of Analytics, Department of Industrial & System Engineering, Aug. 2019 – May. 2021(expected) University of Southern California, United States, Major GPA: 4.0/4.0 Related Coursework: Predictive Analytics (Machine Learning), Performance Analysis Using Markov Models (Stochastic Process), Data Mining

Bachelor of Mathematics & Science, Information and Computing Science, Sep. 2015 – Jun. 2019 Shanghai Normal University, China, Major GPA: 3.6/4.0, Rank 1st/23 TECHNICAL SKILLS

Programming language: Python (pandas, numpy, scikit-learn, matplotlib), SQL, R, C++, Java, MATLAB, SAS, C Tools / Platform: Windows, Linux, Tensorflow, keras, Google Cloud, Google Colab, A/B test Languages: Mandarin (Native), Shanghai Dialect (Native), English (Fluent) WORK EXPERIENCE

Data Scientist Intern - Shanghai Paraview Software Co.,Ltd, China Sep. 2018 – Oct. 2018

Designed a binomial distribution sampling method with large data (n=50k, p=0.01) with R without packages using ‘Binomial random variate generation based on ratio of uniforms, second version’ (BRUS) algorithm

Optimized run-time of BRUS sampling in 1.7ms to increase efficiency of software risk calculation model by 20% PROJECTS

Kaggle Competition: TensorFlow 2.0 Question Answering Dec. 2019 – Feb. 2020

Scheduled and studied NLP course CS224n to consolidate knowledge of BERT to better fit competition

Collaborated with 2 team members to implement ‘BERT joint baseline for natural question’ in Question Answering problem with TensorFlow 2.0 to improve metric --- micro F1 from 0.15 to 0.60, ranked top 11% Bronze Medal – Kaggle Competition: NFL Big Data Bowl Oct. 2019 – Dec. 2019

Extracted, cleaned, aggregated and manipulated data with python from 509k rows data files, handling missing values and merging data to consolidate information to allow users to understand

Created and selected 63 features related to rushers and games by importance, modifying NBA Draymond metric to create a new feature fit to NFL, boosting accuracy by 5%

Built a neural network with keras to predict yards gained or lost in every play and reduced loss with metric of continuous ranked probability score to 0.0132, ranked 8%

PUBG Finish Placement Prediction Oct. 2019 – Dec. 2019

Conducted dropped outliers and cheaters’ data based on data visualization, and selected features with LightGBM importance for 124M+ data to reduce time of evaluation

Applied LightGBM libraries with sklearn in python to predict finish placement and reduced Mean Absolute Error to 0.036 LEADERSHIP & INTEREST

Class President – Nov. 2017 – Jun. 2019

Promulgate information from administration to 22 students and provide weekly feedback to advisors

Held more than 10 studying groups during the session, winning two Excellent Student Leader Awards and improving students’ academic performance --- graduation rate to 96%

Interest: Grade 9 trombone player, NBA basketball fan