Aimun Khan

Vocalitics

My senior year of UT, a group of friends and I developed prototype for a startup. Vocalitics is an interactive meeting analytics tool that draws insights and takes notes on various aspects of a meeting. These insights include a comparison of the speech time of each participant using voice recognition, the dominant sentiment(s) of the meeting, and the sentiments of the meeting as they change over time. These insights are displayed statistically and graphically on the web-app. Additionally, Vocalitics provides note-taking functionality by performing speech to text translations, which renders a formatted and speaker-labeled transcript; automatically scheduling action items on Google Calendar by detecting key phrases throughout the meeting; and displaying the most frequently spoken words in a word cloud. By delivering all of these features into one seamless platform, the Vocalitics team hopes to drastically improve the quality of everyday corporate meetings.

Terminal Live Winner September 2019

Citadel LLC & Citadel Securities partnered with Correlation One to host an invitation-only competition at UT. Correlation One designed a two-player tower defense game in which both players use limited resources to protect their towers while attacking their opponent's. Teams of students spent a day designing algorithms for the game, whose specific rules were unveiled the day of the competition, with the algorithms battling head-to-head in a tournament at the end of the day. My team of three took first place in the tournament. This was a very exciting competition that leveraged both my computer science and math backgrounds simultaneously.

Basketball Predictive Analytics

Some friends and I found a dataset that contained details of every individual shot taken in the NBA during the 2014-2015 regular season, including the distance from the basket, the closest defender, and the result of the shot. After extensive cleaning of this dataset, we created some new aggregate datasets from this individual shot data and analyzed how players match up against each other. We created a tool to visualize these matchups. Finally, we used this data to explore the hot hand phenomenon.

In addition to Scikit-learn and Pandas, this project used D3 to visualize our findings.

Music Taste Prediction Algorithm

I found a 280 GB dataset that contained features of a million songs. A team of my classmates and I did extensive feature engineering to reduce the dataset down to a manageable size. We then designed and implemented an unsupervised learning algorithm that takes a list of songs that an individual likes as input and returns new songs that the algorithms predicts the user would enjoy listening to.

We imported the data to a virtual machine on AWS. This project involved use of Pandas, Numpy, and Scikit-learn.

HackTX 2018 winner - Multi-Factor Voice Authentication Client

At HackTX, my team won first place in the Charles Schwab Security Challenge, as well as third place grand prize. Our goal was to implement a passwordless alternative to authentication. Our solution was to design a multi-factor voice authentication client. Our tool prompts the user to speak the answer to a security question, and then verifies the identity of the speaker using the recorded audio as well as the content of the answer. A prototype of this tool was demo'd during finals of the hackathon.

We used FireBase, Azure, and Tensorflow to implement this tool.

Cloud Computing Research Paper

As part of a research project, a team of students conducted research on the significance of cloud computing. Our goal was to communicate the importance of cloud computing as a disruptive technology to both technical and non-technical audiences. We compiled this research into a paper detailing the technical description of cloud computing as well as its impact on technology and society. This paper were presented to a group of UT undergraduate students.

Inter-App Communication Research Paper

As part of a graduate class I took at UT, Mobile Computing, I wrote a research paper with a group of grad students. In this paper, we explored some of the differences in implementations of inter-app communication in iOS and Android development. Most research in the area of mobile computing defaults to Android as its scope of research, so we wanted to explore the advantages of including the iOS platform in these areas of research.

Freetail Hackers Spring Hackathon 2nd place - Wholesome Studio

We won second place at this Hackathon by making an anti-anxiety web app. Our goal was to create a tool for stressed people to use to calm down by giving them something to focus on with calming audio and visuals. For this web app, we created all of the audio from scratch and used paper.js and howler.js to implement the visuals and connect the audio to these shapes.

Some of the challenges in working on this project including working with unfamiliar JavaScript libraries and finishing an entire deployable application during an eight hour timeframe.

Web Scraping Registration Tool

For my own personal use, I designed and deployed a small Python app to text me via Twilio whenever a closed class that I wanted to register for opened up. This tool used Selenium and Beautiful Soup to scrape the UT registration page, using the Cryptography library to securely log in. This script was deployed on AWS in a Docker container.

HackTX 2017 - Fitbit Heart Monitor

At my first Hackathon, my team designed a Fitbit app and web app to communicate and text an emergency contact the location of the user whenever the fitbit detected a sharp increase in heart rate. We interfaced sensors on a Fitbit Ionic to poll heart rate data from the fitbit and store it on the web app where users could set up notifications. As my first Hackathon, I learned a lot about quickly designing and implementing solutions to a problem using unfamiliar tools. I had never programmed a fitbit app or used JavaScript before, so I had to quickly pick up these skills at the hackathon.

Embedded Systems Video Game

A friend and I designed and programmed an interrupt-based Mario style video game on TI TM4C microcontroller. The game was coded in C, and we created all of the pixel art for the game. We Implemented logic for pressure-sensitive jumping (allowing the length of the jump to be changed by the length of the button press based on a simple physics engine) and parallax background.