Codes

Crafting ideas with logic

CIT Flow

A visualization system for elevators in the Brown CIT

  • Installed Raspberry Pi boards near the elevator to capture the elevator's state
  • Developed an original algorithm to detect the current floor of an elevator, based on computer vision concepts
  • Created a SQL database to store the floor data for future traffic analysis

Python/SQL/Raspberry Pi/OpenCV

Tensona

Implemented a persona-based neural conversation model to handle the issue of speaker inconsistency in natural response generation, as proposed by Jiwei Li et al.


  • Constructed a four-layered Seq2Seq model that contains both the Speaker and Speaker-Addressee Model, as described in the paper
  • Trained the model on datasets consisting of daily multi-turn dialogues and scripts from Friends
  • Achieved a decrease in perplexity of ~7% in comparison to a standard LSTM model

Python/Tensorflow/Keras/Deep Learning/Language Models

Othello

Recreated the board game Othello (Reversi) in Java, adding options of playing against computer AI players with 3 levels of difficulty


  • Utilized object-oriented programming to logically organize different components of the game
  • Designed custom graphical user interface and animations with JavaFX libraries
  • Implemented the MiniMax Algorithm to enable the computer AI to find the next optimal moves corresponding to their selected levels

Java/JavaFX/OOP/AI

Recycle@Brown

Web app in collaboration with Brown’s Sustainability Office that educates users regarding waste-sorting on campus

  • Spearheaded user research with 5 students & staff
  • Designed user interface in accordance with Brown’s visual identity palette
  • Developed responsive front-end interfaces using React.js

Html/CSS/React.js

Shell

Unix shell terminal coded from scratch, capable of:

  • Parsing user input string to determine the appropriate responses
  • Running multiple processes concurrently in the background and foreground
  • Using C system calls to perform file redirection, handle signalling, and run built-in terminal commands

C/Computer systems/String parsing/Job & signal handling

Movie Success

Investigated how certain aspects of a movie affected its revenues & ratings

  • Merged IMDB, TMDB, and World Bank databases and organized & cleaned the resulting SQL database
  • Developed hypothesis regarding how the budget of a movie & the GDP of its production country correlated with its revenues & ratings
  • Tested hypothesis and visualized & analyzed results through multiple regression with Python’s statsmodels

Python/SQL/Data cleaning/Stats


Shell

Unix shell terminal coded from scratch, capable of:

  • Parsing user input string to determine the appropriate responses
  • Running multiple processes concurrently in the background and foreground
  • Using C system calls to perform file redirection, handle signalling, and run built-in terminal commands

C/Computer systems/String parsing/Job & signal handling


Tensona

Implemented a persona-based neural conversation model to handle the issue of speaker inconsistency in natural response generation, as proposed by Jiwei Li et al

  • Constructed a four-layered Seq2Seq model that contains both the Speaker and Speaker-Addressee Model, as described in the paper
  • Trained the model on datasets consisting of daily multi-turn dialogues and scripts from Friends
  • Achieved a decrease in perplexity of ~7% in comparison to a standard LSTM model

Python/Tensorflow/Keras/Deep Learning/Language Models


CIT Flow

A visualization system for elevators in the Brown CIT Building

  • Installed Raspberry Pi boards near the elevator to capture the elevator's state
  • Developed an original algorithm to detect the current floor of an elevator, based on computer vision concepts
  • Created a SQL database to store the floor data for future traffic analysis

Python/SQL/Raspberry Pi/OpenCV


Othello

Recreated the board game Othello (Reversi) in Java, adding options of playing against computer AI players with 3 levels of difficulty

  • Utilized object-oriented programming to logically organize different components of the game
  • Designed custom graphical user interface and animations with JavaFX libraries
  • Implemented the MiniMax Algorithm to enable the computer AI to find the next optimal moves corresponding to their selected levels

Java/JavaFX/OOP/AI


Movie Success

Investigated how certain aspects of a movie affected its revenues & ratings

  • Merged IMDB, TMDB, and World Bank databases and organized & cleaned the resulting SQL database
  • Developed hypothesis regarding how the budget of a movie & the GDP of its production country correlated with its revenues & ratings
  • Tested hypothesis and visualized & analyzed results through multiple regression with Python’s statsmodels

Python/SQL/Data cleaning/Stats


Recycle@Brown

Web app in collaboration with Brown’s Sustainability Office that educates users regarding waste-sorting on campus

  • Spearheaded user research with 5 students & staff
  • Designed user interface in accordance with Brown’s visual identity palette
  • Developed responsive front-end interfaces using React.js

Html/CSS/React.js