Michelle Wang

Staybooking

https://staging.d3q2arguljputk.amplifyapp.com

An online stay rental application based on React and Spring Boot

• Designed and built a responsive web application using React; Bootstrapped the development with Ant Design.
• Developed and maintained RESTful APIs and Spring Boot-based microservices, enabling users to create, search, reserve, and manage stay listings.
• Leveraged PostgreSQL, PostGIS, and Hibernate (JPA) for storing user data and geospatial queries; used Google Cloud Storage for managing uploaded media files.
• Implemented stateless and scalable user access using Spring Security with JWT token-based authentication.
• Deployed the backend to Google Cloud Run, and the frontend to AWS Amplify, with CI/CD integration using Gradle and Git Action for automated builds and delivery.

Twitch+

https://mcrawjf52p.us-east-2.awsapprunner.com/

A personalized Twitch resources recommendation engine

• Built a web page with user-friendly UX using React and Ant Design; Stored game data in MySQL database hosted on AWS RDS; Conducted CRUD operations with Spring Data JDBC.
• Integrated the Twitch API using OpenFeign to fetch real-time Twitch content and built RESTful APIs for frontend-backend communication.
• Implemented Spring Security for user authentication and developed a content-based recommendation system using users' favorite collections.
• Optimized performance with Spring Caching + Caffeine and ensured reliability through unit testing.
• Dockerized the application and deployed it seamlessly on AWS App Runner for public access.

Next AI

https://staging.d3szzyszv8ohsw.amplifyapp.com/

A full-stack web-based Q&A AI agent for PDF document queries

• Created an interactive conversational UI leveraging React and Ant Design, enabling users to effortlessly upload and interact with PDF documents in real-time.
• Implemented RESTful APIs via Express and Node.js and optimized for high-performance request handling.
• Utilized an in-memory vector store to cache generated embeddings for efficient retrieval.
• Integrated OpenAI's GPT-3.5 Turbo API and LangChain technologies, resulting in a sophisticated AI Agent capable of document loading, splitting, storage, retrieval, and output functionalities.

Global Terrorism Analysis

A PostgreSQL-powered analytics system for identifying terrorism patterns

• Cleaned dataset with 180,000+ entries and 135 variables using Python; designed a relational schema in 3NF with ER diagram including all integrity constraints and triggers.
• Extracted 23 key variables related to the project scope from original dataset, performed data cleaning and aggregation to transform the selected data, and staged the data on destination database by creating a connecting engine with PostgreSQL server.
• Built an interactive dashboard visualizing targeted terrorism KPIs using Metabase, and summarized findings in a report presented to the United Nations Office of Counter-Terrorism.

YouTubeTrends

A Python-based analytics project uncovering factors behind YouTube viewership

• Implemented data-cleaning process on 40,000+ observations and 17 variables including both structured and unstructured data; transforming fragmented datasets into a unified dataset for streamlined analysis.
• Utilized Pandas and Matplotlib libraries to visualize data insights; plotted correlation matrix and visualized the result using heatmaps to identify statistical relationships between numerical and Boolean variables.
• Applied sentiment analysis on video titles, genres, and description keywords to identify and monitor viewer sentiment qualities and computed word-clouds to represent frequency counts and importance of words.