- Construct and initiate the data collection process to set up the recommendation system- Develop the recommendation system to predict the preference of a user and solved the cold-start problems.- Deployed MySQL to prepare, design, and develop the backend database in order to complete the connection between the frontend and the backend as well as between the backend and the data models.- Designed a complete process of doing data collection and data processing as well as developing the database with the backend.- Currently in the process of building several machine learning models to compare image similarities.

- Used Optical Characteristic Recognition(OCR), PyPDF2, and tabula as well as developing a pipeline to convert more than 6000 pages data in PDF format into CSV format.- Web-scraped thousands of customer data from their website by selenium, bs4, requests, regular expression and other techniques.- Developed the customer persona of their company and constructed the marketing plan for them by using statistical testing as well as tools such as data manipulation, data cleaning, and data visualization.- Advised and reported on how to develop their pricing platform for a new insurance company they plan to set up soon.

- Deployed Python with Tensorflow to train multilayer neural networks, recurrent neural networks and other deep learning algorithms in order to perform language modeling and Natural Language Processing, preparing for the AI lab.- Joined the AI(Artificial Intelligence) Lab and attended the weekly discussions, seminars and researches of Document Understanding and Optical Character Recognition of Natural Language Processing. Preliminarily involved in text and character recognition algorithms such as PlugNet and Super-Resolution Unit.

- Deployed MySQL and SparkSQL for data collection, ETL, and EDA, and created datasets for difference business topics.- Utilized Python with AI packages(NLTK) for vectorizations and word bedding, focusing on topics related to 5G, data networks, and Wi-Fi internet services.- Used Vader score to calculate tentative ratings for comments, and produced sentiment analysis using NLP.- Applied models such as Naive Bayes, Decision Trees, Neural Networks in order to predict scores with accuracy around 85%.- Adopted R to perform time series on the number of hits on related topics, gave predictions, and provided a range of error.

- Investigated and reported VPN users’ operating system, Client Groups, Clients in different ISPs, Clients in different regions and countries with users’ permissions on campus with the web-based cloud storage platform, Splunk Inc, R, and Python.- Automated data collection and data visualization process of VPN information to implement real-time tracking and analysis.- Cooperated with the computer information security team, advised on deploying targeted services.

- Led the daily training and tactical drills based on each trainees' strengths and weaknesses- Assisted targeted tactics development for each game.

- Cooperated with the business assistant team, provided advice, and determined the process of data collection and collation.- Investigated 7,000+ past sales data points by R and Excel, performed data visualization techniques to demonstrate the sales and revenue trends and preliminary regularity in months, quarters, districts, and industrial fields.- Implemented data manipulations, linear regression model and machine learning techniques to propose quotes to maximize the profit margins for each device category. Based on the particularity of this industry and the demand-supply integrations, manually constructed the business bidding documents and optimized the production lines directed by the manager. - Conduced to the development of business marketing strategies and annual business report including the prediction of the revenue for the first half-year of 2018.