Dzung V. Pham

Hello, my name is Dzung Pham and I'm a 2nd-year Computer Science PhD student at University of Massachusetts Amherst, where I'm researching secure, private, and trustworthy AI/ML and systems with Prof. Amir Houmansadr.

Previously, I worked as a Software Engineer at Meta, Inc (formerly known as Facebook), where I applied ML to protect users of Facebook Marketplace from fraud, scam and harassment.

I graduated from Williams College in 2020 with a degree in Computer Science and Statistics. I was advised by Prof. Richard De Veaux for my senior thesis. I also had the pleasure of working closely with Prof. Daniel Barowy and Prof. Duane Bailey.

Email  /  CV (2023)  /  Google Scholar  /  Github  /  LinkedIn

profile photo
RAIFLE: Reconstruction Attack on Interaction-based Federated Learning
Dzung Pham, Shreyas Kulkarni, Amir Houmansadr
Under review
arXiv / github

Proposed and analyzed a novel active attack in interaction-based FL scenarios, particularly federated recommendation/learning-to-rank.

Evaluating ProDirect Manipulation in Hour of Code
Quan Do, Kiersten Campbell, Emmie Hine, Dzung Pham, Alex Taylor, Iris Howley, Daniel W. Barowy
SPLASH-E, 2019
project page / paper

Developed SWELL - a novel language for teaching beginners how to code via a bidirectional link between code and outputs.


Below are some fun projects that I worked on during my undergrad:

FallNet - Independent Winter Study Research Project, Williams College

I trained a two-stream Convolutional Neural Net to detect people falling from video input, with transfer learning from MobileNetV2 and Motion History Image in place of the optical flow stream. This project won the 2019 Ward Prize for Best Project in Computer Science at Williams College.

Facial Emotion Recognition - CSCI 373, Williams College

I trained a Convolutional Neural Net to detect facial emotion from images, then built a program to do so in real-time via webcam. This was my final project for Artificial Intelligence. Trivia: the emotion 'Disgusted' was the hardest to detect, and also the hardest to recreate.

CUDA C Implementation of the Lucas-Kanade Algorithm - CSCI 338, Williams College

I implemented the Lucas-Kanade optical flow algorithm for GPU using CUDA C. This was my final project for Parallel Processing. I believe this was the only project for which I got an honorary A+.

Rhythm Bracer - eTextile, Winter Study 2018, Williams College

I created an Arduino-powered cloth bracer that can record and replay tapping patterns. The most important thing I learned from this project was how to sew and embroider, probably.

Random stuff

I have been playing the classical guitar unprofessionally since grade 9, with a bit of flamenco and fingerstyle here and there. The proudest piece in my repertoire is Koyunbaba by Carlo Domeniconi. I had the honor of studying under Robert Phelps - a true professional guitarist.

Recently I got addicted to chess (following the Queen's Gambit craze). You can find me on

This website was based on here.