Qin (Aria) Wu

Software Developer Intern | We Care STL, St. Louis, MO 2021/05 - Present

https://www.wecarestl.org/

• Maintained, updated, and troubleshoot organizational React website from scratch including microsites, applications and forms
• Currently collaborate with the team of 5 to design features of a brand-new website built with Laravel PHP framework and MySQL database

GoParks Web Application 2021/01 - 2021/04

https://goparks.herokuapp.com/

• Designed and implemented GoParks website that supports CRUD operations for users, parks and reviews depending on a user’s authorization/ownership.




• Utilized RESTful architecture with Bootstrap 5 for styling; developed the backend using Node.js and Express framework; implemented MongoDB Atlas, a cloud-based NoSQL database, to store data.

• Integrated Mapbox to provide parks’ information and link on map based on its location.

• Designed flash messages handling error and success results to provide users with feedback.

• Hashed password using a cryptographic salt by Passport.js to enhance the authentication mechanism.
• Deployed GoParks app through Heroku.

MovieBox React using Node Backend 2020/09 - 2020/11

https://github.com/oneariaaa/MovieBox

• Built with React and Bootstrap5 on the front-end implementing pagination, categories, searching, and sorting of objects, etc.




Handled routes and redirects for different URLs form submissions; Form validation using JOI library.
Authentication and Authorization using JSON Web Tokens; Integrated Sentry to track and be alerted to HTTP errors; Error logging using Winston library.
REST API is used for data exchange between the front-end and the backend.

Data visualization and predictive analysis of electronic game sales 2020/10 - 2020/11

• Preprocess dataset by missing value filling and standardization. Execute exploratory data analysis.
• Visualize the data to show the commercial value of the data, such as obtaining the most popular game types in the world, the publisher with the most games in the world and the game platform with the most sales in the world.
• Based on the information of game type, publisher and platform, implementing one-hot coding and establish a linear regression model to predict the sales volume of the game.

Regression and classification analysis on UCI wine dataset 2020/03 - 2020/04

https://github.com/oneariaaa/data_science_course

• Preprocess dataset by data cleaning, categorical feature transformation, missing value filling and standardization, etc.
• Visually analyze the data to determine the analysis steps and methods.
• Execute linear regression, Lasso regression, and random forest to analyze the 14 chemical information variables of wine, obtain the model to calculate the predicted value; evaluate the model through the average absolute error.
• Implement logistic regression, K-nearest neighbor algorithm and self-built model to learn the training set and evaluate the model through the average absolute error.
• Refine the model by finding the best parameters using grid search (GridSearchCV). The final accuracy rate reached 99.8%.

Computer vision: count the paperclips 2020/03 - 2020/04

https://www.kaggle.com/c/applications-of-deep-learningwustl-spring-2020b/overview

• Preprocess the 50000 images with papercilps by ImageDataGenerator function from Keras.
• Customize a CNN model to train and learn the feature of images.
• Implement RMSE to measure the difference between the predicted number of paperclips and the actual number of paperclips.

MINST Handwritten Dataset 2019/10 - 2019/11

• Apply Keras package to deploy the convolutional neural network; customize a VAE model; Calculate the mean and variance of each number k; Construct normal distributions and perform sampling reconstruction.
• Customize a convolutional layer to calculate the loss between the input image and the decoded image.
• Apply cosine similarity comparison method to compare the similarity of the images; Modify the number of dimensions and layers to generate the best parameters of the model.



• Plot 3-D normal distributions to show both Vk (V0) and Vk’ (the difference between V0~9 and V1~9).

• Change the mean value by increasing the i-fold delta value, and compare the similarity of the two sets of data again to determine the best parameters.