My Projects
Sourceformer: Tool Integration into Transformer Large Language Models Through Source Code
This was a REU (Research Experiences for Undergrads) program at East Carolina University. I spent my almost my entire summer in North Carolina proposing and implementing this AI research paper. My mentor was Dr. Nic Herndon.
I had such a great time, and I am super grateful for this opportunity. The mentors and my student colleagues were such a blast to be around whether we were talking AI or just chatting about life.
Paper | Difficulty: 10/10 | 2023
My creativity experiment
Every day during my spring semester I headed to The Den at 5:30am. If anyone asked, it was my office. But in reality it was a 10x10 storage unit I had rented out. Inside was a desk, a chair, and a small heater.
I wanted to explore what creativity was without any preconceived notions about it. For that I needed a quiet, consistent, and safe place to focus.
Difficulty: 9/10 | 2023
Modified Dibels (AI model)
This was the second half of my research under Dr. Jake Qualls at Astate. My job was to build the font-readability CNN which takes an image of a font character as input and outputs a readability score from 0 to 1. Collecting these readability scores was part 1: The website and testing phase of this research project. We started with the Attr2font GAN, and the idea was to run our font-readability CNN in parallel with the discriminator in hopes that this would generate highly readable fonts for dyslexic.
Code | Paper | Difficulty: 7/10 | 2022
Modified Dibels (website)
This was the first half of my research under Dr. Jake Qualls at Astate. The site is a modified version of two common dyslexic literacy tests. We are using the data collected from the dyslexic students who take these tests to train an Ai model. Our hypothesis is that this model will create the most readable font for dyslexic readers. I used the Django and PostgreSQL to build it.
Code | Difficulty: 10/10 | 2022
Avant-Gardes: Rick Rolled
This was a NFT project I did back when the Bored Ape Yacht Club just came out. Feel free to check it out.
Code | Difficulty: 10/10 | 2021
CS50x - parody (CS50x final project)
I loved CS50x so much, and I wanted a way to quickly look back on the problem sets.
I scraped their website and made a Flask app out of it. There are sooo many CS50x inside jokes on this website if you look hard enough ;)
To anyone who has taken this course, I think you will find this really funny!!
Code | Difficulty: 8/10 | 2020
A structured programming alternative
Structured Programming is the first computer science course for most students at Astate. However, I thought there was alot we could do to improve it.
I spent a month thinking and writing out all of my ideas.
These ideas haven't caused much change just yet. But I still believe in them, so we will see.
Difficulty: 7/10 | 2022
Three weeks of machine learning
I hadn't committed to a personal project in a while. So I wanted to brush up on some machine learning, and just get something completed. The first week I build a simple feed forward neural network by hand. The second week I did a CNN by hand. The third week I did a transformer, but I definitely used some libraries to help me lol.
Difficulty: 6/10 | 2022
Richey's Rivalry
This was a very fufilling fundraising project for a friend, and I'm proud of what I was able to coordinate. That's all I'll say about this here.
Difficulty: 6/10 | 2020
Reading smart contracts
I wanted to better understand Solidity, and hopefully think of a new web3 application. I mainly dove into SushiSwap and WonderlandDAO's contracts; although I poked around in some others. It was so cool being able to talk to the contract creators over the internet. For example BoringCrypto answered a few questions about his really clever BentoBox contract.
Difficulty: 5/10 | 2021
Tab Predictor
I thought it would be fun to build a basic model that predicts what tab you will open when you first open google chrome. In retrospect I should have used a different model structure, (RNN or transformer) and I only got the accuracy up to 60% on the test set.
Difficulty: 5/10 | 2020
Flamethrower
I got tired of writing code. I did this instead.
Difficulty: 3/10 | 2020
Personal Podcast
This is the basic idea of what this is. Here Rick Rubin is talking about what a personal podcast is.
So far I've only done one personal podcast with one of my oldest friends. I thought it was a lot of fun and I learned a bunch. I might do more. We will see.
Difficulty: 2/10 | 2023-01
Vex EDR Clawbot
I had to make a guide for the EDR Clawbot during my senior year in highschool. Not hard but pretty fun.
Difficulty: 2/10 | 2020