Data Scientist Analyst

.

ZINIU YU

Irvine, California ***** 949-***-**** *******@***.*** linkedin.com/in/ziniu-yu-2396291a6 kaggle.com/unizy22 S U M M A R Y

• Current Master of Data Science student with internship experiences as an analyst and data scientist.

• Strong skills in using Web Crawlers to collect data, applying SQL to manage data, and building Statistical or ML models to solve problems.

T E C H N I C A L S K I L L S

Core Domain Expertise: Data Analysis, Statistical Modeling, Machine Learning, Data Mining Tools: Python3, Web Crawler, Hive SQL, Postgresql, Jupyter, Pytorch, Pandas, Numpy, Keras, Keras Tuner E D U C A T I O N

University of California, Irvine, California

Master of Data Science, GPA: 3.90/4.00 Expected Dec. 2022 Harbin Engineering University, Harbin, China

B.S., Mathematics and Applied Mathematics, GPA: 3.62/4.00 June. 2020 E X P E R I E N C E

XIAOMI TECHNOLOGY. Wuhan, China

Data Scientist Intern, Big Data Department Oct. 2019 – Jan. 2020

• Built a word2vec word vector NLP model to calculate the similarity between finance words. Expanded 5,000 words, and enhanced the recall of the tagging process by 3%.

• Deployed a Web Crawler to collect music entities, added more than 10,000 entities, and 87% of them were utilized for tagging users to make advertising more precise.

• Operated hive SQL to query music data, analyzed the popularity of them with Pandas, and helped increased DAU by 5000.

• Coded in Scala, helped improve the data mining logic, and improved the accuracy of tagging process to 90%. WUHAN BUREAU OF STATISTICS. Wuhan, China

Analyst Intern Aug. 2019 – Sept. 2019

• Estimated the total factor productivity (TFP) leveraging Cobb–Douglas production function. Helped researchers understand the impact of technological innovation on Wuhan’s economy from a quantitative perspective.

• Applied the Time Series Regression to predict the potential economic growth of Wuhan, and contributed to a theoretical paper about the influence of technology development on the economy. P R O J E C T S

Real-time gesture recognition, ● Python3 ● Pytorch ● OpenCV ● Numpy ● Pandas Nov. 2021 – Dec. 2021 Real-time gesture-classifier using neural networks(SSD & Resnet) on Pytorch. (https://youtu.be/Kx9p1sGUAGg)

• Used EgoHands and COCO-Hands datasets to train the SSD300 model, connected it to the web camera by OpenCV, and developed a hand detector with 18 FPS.

• Performed transfer training on the ResNet-18 model with accuracy of 98%. Connected it to the hand detector and constructed a real-time gesture classifier with 12 FPS. Jane Street Market Prediction, ● Python3 ● Keras ● Keras Tuner ● Numpy ● Pandas Dec. 2019 – Feb. 2021 Kaggle competition that uses attributes to decide whether a trade should be made.

• Cleaned the data that has little impact on decision-making, filtered the features bias with Autoencoder, adjusted the hyperparameters with Keras Tuner, and boosted the score by 3000.

• Constructed an MLP model by Keras to determine whether the transaction should proceed, ranked top 17% on the public leaderboard, and defeated 3176 teams on the private leaderboard. Traffic Mode Recognition, ● Python3 ● Sklearn ● Numpy ● Pandas Jan. 2019 –Feb. 2019 A classifier based on the random forest model to distinguish traffic mode.

• Used the genetic algorithm on Matlab to calculate the approximate signal function, found a suitable function.

• Constructed a random forest model with Sklearn, and got a classifier of traffic mode whose accuracy and recall are both over 90%.

Contact this candidate