Full stack web development

Developed an online learning management system for technical training and job placement following MVC architecture.

SQL database

Designed database schema, managed platform data in MS SQL and connected to backend server through SQLAlchemy

Python backend

Implemented backend data server using Python Flask and built RESTFUL APIs for front end clients.

User Interface

Built modularized user interface with react & Material UI and handled data flow with Redux State container.

Modern Deployment

Deployed on AWS EC2 with Nignx and Gunicorn, configured reverse proxy load balancer which improved QPS by 37%.

Natural Language Processing

Utilized cutting edge NLP tools and research to analyze the persuasivenss of debate under guidance of prof. Jingbo Shang.

Full stack web development

Used Flask, AWS and Bootstrap to build an online platform for researchers and debater to interact with our engine.

Feature Extraction

Tailored GloVe embedding and attention mechanism to extract relevant information from both sides of debate.

Recurrent Neural Net

Implemented LSTM to capture long-term depency within debate speeches and to gain insight about logical structure.

Incubator and Community

Led both external and internal departments of the largest on-campus incubator for 12 student CS/DS projects with over 110 members.

Connection

Outreached student orgs, professors and industy partners (including Qualcomm, TuSimple, ServiceNow and many more) and raised over 5k in funding.

Business Development

Brainstormed project ideas, developed business plans and prepared pitch presentations for quarterly projects.

Operations and Event Organization

Organized weekly meetings, on- and off-line workshops, social events and quarterly Demo Day.

Market Making Agent

Researched on competitive market making via deep reinforcement learning under guidance of prof. Sicun Gao.

High-fidelity Simulation

Built realistic and data-driven simulation of a limit order book using historical data and sequential Bayesian methods.

Deep Reinforcement Learning

Combined function approximation and constraints satisfaction in the design of action, state, and reward representations.

Fine-tuned Model

Design customized Gym environment for deep reinforcement learning and represent stock features using RNN.

Data Structure and Algorithm

Tutored DSC30 data structure and algorithm under the guidance of prof. Marina Langlois.

Recommender System

Tutored CSE158 data mining and recommender systems under the guidance of prof. Julian McAuley.

Tutor Hours

Performed individual and group tutoring for students and helped them with concepts, practice, and assignment.

Mini-sessions

Organized mini sessions to review essential concepts covered in lectures and prompted active discussions.

Kaggle Competition

Built category classifier on Amazon and Yelp's dataset (~200,000), with accuracy ranking 2nd out of 355 on Kaggle.

Feature Extraction

Designed the classifier using Scikit-Learn's linear SVM, extracting features from review text and purchase history.

Customized Algorithm

Developed a collaborative filtering algorithm for purchase prediction on Amazon's data, ranking top 12% on Kaggle.