About Me

Mai Elkady

I am a fifth year computer science PhD student at Purdue University in West Lafayette, IN. I got my masters in Computer Science from the same university in May 2018. I am studying under the supervision of Prof. David I.Inouye and Prof. Petros Drineas. My areas of research interest are Generative Models, Normalizing flows, Diffusion Models, Computational Biology and Bioinfomatics. I am originally from Cairo, Egypt, and I did not realize how much I miss the sun except after spending a winter in West Lafayette, IN.

My Education

Purdue University, West Lafayette, IN, US

GPA: 3.8, Advisors: David I.Inouye & Petros Drineas, Honors: Graduate Teaching Award for Spring 2020

Aug. 2018 - Present
PhD in Computer Science

Purdue University, West Lafayette, IN, US

GPA: 3.8, Honors: Fulbright Scholarship

Aug. 2016 - May 2018
MS in Computer Science

Ain Shams University, Cairo, Egypt

GPA: 3.58

Aug. 2007 - June 2012
BS in Communications System Engineering

My Skills

My Publications

Discrete Tree Flows via Tree-Structured Permutations

Authors: Mai Elkady, Jim Lim, David I. Inouye. Accepted in: ICML, July 2022

Discrete Tree Flows via Tree-Structured Permutations

Authors: Mai Elkady, Jim Lim, David I. Inouye. Accepted in: ICML Workshop on Invertible Neural Networks, Normalizing Flows, and Explicit Likelihood Models (INNF+), July 2021

Submissions to a COVID-19 Data Science Challenge: the role of skills and platform engagement

Authors: Sabine Brunswicker, Mai Elkady, Feny Patel. Accepted in: ACM Collective Intelligence Conference, June 2021

TeraPCA: A fast and scalable software package to study genetic variation in tera-scale genotypes

Authors: A. Bose, V. Kalantzis, E. Kontopoulou, M. Elkady, P. Paschou, P. Drineas. Accepted in: Bioinformatics


My Projects

Predicting user participation in Data Science Challenges

As part of my work as a fellow at the Research Center for Open Digital Innovation (RCDOI), I analyzed participants data in COVID-19 data science competitions on the Ironhacks platform to gather interesting observations, and determine the important factors that best predicts whether a participant will submit or not. The results were presented at ACM Collective Intelligence Conference 2021

IronHacks COVID-19 Data Science Challenge

Participated and won third place in the Ironhacks COVID-19 Data Science Challenge, where the task was to predict the weekly foot traffic at merchants in Indiana in order to understand the COVID-19 impact and risk. To solve this problem I used Python to train a ridge regression model that was able to obtain good results in predicting the foot traffic at various stores in Indiana.

Flower Species Identification

Employed a DenseNet pre-trained Convolutional Neural Network model to train an image classifier to identify 102 different species of flowers. The code was written in Python and used PyTorch for deep learning, and the training was done utilizing GPUs on Google Colab. The project was then deployed as a webapp using Flask on herokuapp

Synthetic Genotype Data Simulator

As part of a team, implemented a data simulator in C/C++ that generates synthetic genotype data using the Pritchard-Stephens-Donnelly (PSD) model.

Predicting Pulp Fiction Lovers

As part of a class Kaggle competition, tried several Machine learning approaches, and coded them in R and Python, to predict whether users will like the movie Pulp Fiction given their previous movie ratings.