Profile Website

The website you're viewing now! The purpose of this project is to create and design my own platform to display my portfolio as a developer and help inter-link all of my relevant profiles online so that people can reach me and learn about me more easily. Tech used:

  • HTML5
  • CSS3
  • Javscript
  • JQuery

Web Design/Dev

RBM on MNIST

Learning project following my study of RBMs in Geoffrey Hinton's Coursera course on neural networks. Trains a generative model to learn a probability distribution through its hidden states to generate binary reconstruction vectors of the MNIST dataset. Uses Gibbs Sampling for MCMC sampling.

  • Tensorflow
  • RBM Neural Network architecture

Machine Learning, Python

Satellite Road Extraction (In Progress)

Uses GIS and Web Map Service to build a dataset of satellite/aerial images within a bounding box of coordinates. Will investigate the results of roadmap generation via a Residual U-Net Neural Network. I'll also try to investigate and compare the results of different architectures.

  • Tensorflow
  • Neural Networks - Res U-Net

Machine Learning, Python, In-progress

RWD Practice

A sample website without a theme or subject -- it aims to test and demonstrate fundamental ideas in responsive web design. Has a homepage and a gallery page which both adapt to mobile via media queries. The gallery uses jQuery to add a third column on wider displays.

  • Responsive Web Design
  • HTML5, CSS3, Javascript, jQuery
  • Gallery, modals

Web Design/Dev

Graph Theory

In this project, I used C++ to create a graph theory system that allows the construction of graphs using vertex objects, and provides custom-defined algorithms for BFS traversal/search. The implementation seeks to be memory-efficient, and the headers used are as follows:

  • Functional.h -- specifically for reference-wrappers
  • Algorithms.h -- so I don't have to focus on manipulating the STL structures as much
  • STL Data structures: Queue, Lists, Vectors

Dev, C++

Word2Vec

An implementation of Word2Vec via a skip-gram model aiming to learn word embeddings for NLP purposes.

Machine Learning, Python

Match Card Game

Final project done for a CS course that made use of C++. This is a game (simulation) where multiple players may participate and attempt to empty their hands by matching either the rank or suit of the current face-up card.

This project demonstrates understanding of OOP including classes, inheritance, operator overloading, and construction of fundamental data structures from scratch (stack and queue).

Dev, C++

Flight Oracle

This was a team project done for a hackathon. The system consists of a web interface that takes a user's flight destination and time of interest to query the Weather Network's API for relevant weather information. This is then fed into a trained model that predicts that likelihood of a delay exceeding 30 minutes.

I assisted with data gathering, and handled processing/cleansing the dataset as well assembling the prediction model.

Web Design/Dev, Machine Learning, Python

Coursera ML Projects

This repo contains my responses to the assignments given in Andrew Nguyen's Machine Learning course on coursera. All assignments were submitted and received full passing marks.

Machine Learning, MATLAB/Octave

Textadventure

An old textadventure project that I created a long time back. It's set up as a turn-based rpg with four hero types, and same-machine multiplayer. Functionality has been included for interactive gameplay (leveling up, enchanting), but the game doesn't actually have any gameplay beyond creating and naming the players' characters (I had to focus on exams shortly after).

Dev, C++