Columbia University in the City of New York

National Taiwan University

Zhejiang University

Senior Software Engineer at Uber

San Francisco, California, United States

Uber

Senior Software Engineer

- Designed and implemented distributed GBM and logistic regression algorithms, 5 times faster than Spark MLlib•	Used parameter server and OpenMP to create a distributed deep learning framework with automatic resource elasticity•	Developed image classification pipeline based on state-of-the-art neural networks with high accuracy and scalability•	Proposed and constructed efficient and user-friendly Object Detection application using fine-tuned Faster-RCNN model

Petuum, Inc.

Data Scientist

- Designed and deployed MapReduce data preprocessing jobs on Hadoop to extract and transform features from 1 TB raw data•	Applied distributed cluster analysis to build a customer segmentation model using Spark, improving over 130% homogeneity•	Created fraud detection model using doc2vec and GBM on customer claims, gaining 83% match rate with manual labeling 

Cablevision

Data Scientist Intern

- Used OpenCV and convolutional neural networks to build an image emotion recognition application with 67% accuracy•	Utilized bag-of-words, word2vec and classification models to build a sentiment analysis pipeline with 0.74 F1-score  

Research Assistant

Haoyang Chen