Introduction

Hi, my name is Minh Duc Nguyen, and I am a Computer Science and (soon to be) Mathematics major at Tufts University. A little bit about myself: I have been a lover of numbers and algorithms since secondary school, studying in specialized mathematics classes and competing in various national-level competitions. Then, I was introduced to Computer Science in my freshman year of college, and immediately became interested in how expertise in mathematics (and especially in algorithms) can be used to develope ingenious software that better the life of people and facilitate the development of science. If you are also a Math or Computer enthusiast and want to chat, feel free to send me an email or connect with me through LinkedIn!

Academic

Tufts University

2015 - 2019

  • Majors: Computer Science and Mathematics.
  • GPA: 3.96.
  • Honors and Awards: Dean's List (every semester).

Hanoi University of Science - High School for Gifted Students

2012 - 2015

  • Specialized class in Mathematics.
  • GPA: 9.7 (Scale of 10).
  • Honors and Awards: Top 4 (Second Prize) - Vietnam National Mathematics Olympiad (2015)

Experience

Facebook, Inc (Speech Team - Applied Machine Learning Org)

Software Engineer Intern,

Summer 2018
  • Improve the efficiency of the evaluation process in speech recognition model by implementing a pipeline utilizing GPU power
  • Optimize resource utilization of the acoustic model by building an interface to process input data in concurrent batches instead of in sequences

EditShare, LLC

Software Engineer Intern,

Summer 2017
  • Implemented a new file sharing mode for a network-attached storage based on a distributed Linux filesystem
  • Implemented a server API using Twisted Perspective Broker framework to handle Remote Procedure Calls (RPCs)
  • Created a cross-platform Graphical User Interface application using the Qt and PyQt frameworks allowing users to make RPCs to the server API

Tufts University Human - Robotics Interaction Lab

Undergraduate Research Assistant,

Fall 2016 - Spring 2017
  • Implemented the Python version using TensorFlow for the neural field model for real time speech perception project
  • Participated in discussions on how to improve the prediction accuracy of the model

Hanoi University of Science High School for Gifted Students (HUS - HSGS)

Computer Science Olympiad Team Trainer,

Summer 2016
  • Prepared the HUS HSGS Computer Science Olympiad Team, one of the best in Vietnam, for national and international competitions such as the International Olympiad in Informatics
  • Taught how mathematics concepts and theorems such as the Hall’s marriage theorem or the invariant principle can be applied in computer science
  • Introduced the students to various advanced data structures, such as Binary Indexed Tree and Interval Tree, and their applications

Projects

Trailer Nailer

Web application

  • Implemented a web game that allows players to test their movie recognition ability by watching trailers
  • Used MongoDB to store information of roughly 200 movies from 5 different genres, and used Node.js to implement a web server to respond to movie information queries
  • Used YouTube API to embed in-game movie trailer videos
  • Other technologies: Facebook API, Bootstrap
  • Try it out here

Gomoku AI

Python and C++ program

  • Implemented two different Artificial Inteligence (AI) programs that can play the game Gomoku
  • The first AI mirrored my own strategy for Gomoku and made move decision based on the score calculated for each possible move. The score calculation rules were set and adjusted based on how closely the AI follow my expected move.
  • The second AI utilized the Minimax Algorithm to calculate the best possible move under the assumption that the opponent also played an optimal game.
  • Researching to see how a third AI can be implemented using deep learning
  • Take a look here

Neural Field for Speech Recognition

Python program

  • Given the implementation of the neural field model in C++, re-implemented the model in Python using TensorFlow for better scalability and efficiency
  • Doubled the maximum size of the sound data set processable by the model by using sparse tensors
  • Modularized the implementation of the neural network by utilizing tensor slicing and joining, allowing the model to support any given number of neural field layers

Scanned Document Black Edge Remover

C program

  • Used depth first search to identify and delete black edge pixels in scanned images of document
  • Used C, written in Linux environment

Contacts

LinkedIn   Github     Gmail