software
General software products, most of which are web apps or small backend services.
Next Tech
Suggests the next language/framework to learn based task needs. Powered by OpenAI and Vercel edge functions.
ViewPOTG
A short video platform for game clips. Built with Angular, TailwindCSS, Firebase and FFmpeg (WASM).
ViewQ Programming Language
Some day I decided to make my own programming language for data science that combines many nice features of R and Python. It should have no weird syntax, and be vectorized and functional in nature.
ViewInteractive XKCD-styled chart for R
A htmlwidgets package to create interactive XKCD-styled charts in R. This is a wrapper around the javascript chart.xkcd library.
ViewAPI for Game Statistics in the Overwatch League
An API service for querying game and player statistics in the Overwatch League.
ViewJavaScript Notebook
The jupyter notebook for JavaScript and TypeScript to run any code and 3rd party libraries. Bundling and transpiling directly in the brower.
Viewdata-science
Projects involving data analysis and machine learning, typically written in R or Python. Presented as a minimal product as a package, web application, report or API that can be called by others.
Integrating transformers into IOS apps
An ios app for experimenting with using transformers in ios apps, consists of a chatbot with GP3, drawing pad with vision transformer, and a question answering app with DistillBert.
Viewagua: tidymodel's integration to the H2O platform
What I have been working on as the 2022 summer intern at RStudio. agua provides model fitting and tuning infrastructure to let users leverage h2o's computing capabilities while still using tidymodel's syntax.
ViewAirbnb listings in NYC
A shiny app (also an R package) to explore Airbnb listings in NYC. Users can select room location, enter housing descriptions, and get price predictions. If you are into maps, there is also a spatial section analyzing the housing's spatial pattern. There are also a gallery of interactive visualizations for EDA.
ViewETL pipelines for Deutsche Börse Group daily trading data
A PySpark ETL pipeline to process the daily trading data of Deutsche Börse Group and save clean data to AWS S3. Well tested and highly configurable.
ViewModeling titanic survival
Ok I promise you this is not one of Kaggle's introduction to machine learning tutorials. This is a rather pedantic study using the most up-to-date and complete Titanic dataset from encyclopedia-titanica.org to model survival rates. The modeling process includes redundancy analysis, multiple imputation, penalization, bootstrap validation, partial dependence and so on.
View