Nate Weaver

This project was a culmination of my entire degree at CU Boulder! Tenfins Portal is a hub for Flutter developers to connect, share projects, and create together. It was a collaboration with a significant team, where I took on a the role as team lead focused for our AI and channels APIs. The main scope of the project is a messaging and channels system built on a GoLang backend, talking to a PostgreSQL database, and we focused heavily on scalability and maintainability throughout its development. From this, we were able to create a product built on GCP, using Kubernetes to scale the application, and containers deployed to maintain all AI services we built. While our project had very loose leadership from our sponsor, which was a challenge, it allowed me to take the project by the reigns and learn the things I wanted to, as well as step into a leadership role within the team. Not only was I able to improve my comfortability with delegating work and responsibilities, but I was also able to learn a lot technically, from tools like Typescript to Kubernetes. This was a great project to finish my degree with, and I am proud of the work we were able to accomplish despite the challenges we faced!

This project was a final project for my Data Mining class at CU Boulder. This project was worked on with a couple of other friends, and was a great learning experience overall. In this project we used a multitude of data science techniques to analyze the relationship between NCAA and NBA players to see if there was a correlation between performance in college and the NBA as well as determining a success statistic for this relationship. Though this project was split up into a few parts, one of the hardest parts was the collection and cleaning of the data. The data was not very clean, and had to be cleaned and processed to get it into a usable format. Our results gave us a solid statistic to determine success, and did notice some correlation between areas and types of players that were successful there. We did have some disconnect in our ML models, which only got to about 85% accuracy, but overall I think we did a great job with the time we had!

This project was a final project for my Advnced Data Science class at CU Boulder. This project was worked on with my friend Kevin, and was a great learning experience overall. In this project we implemented K-means clustering on the MNIST and Fashion-MNIST datasets and then changed parameters trying to find the best way to intialize centroids, iterate and update centroids. This project was interesting because it was a bit difficult to determine what was actually happening. To really understand the processes, it was important to understand not only the numerical data, but what K-means was doing which incorporates an unusual abstract complexity. Overall, I'm really happy with how this project turned out, I learned a lot and had a lot of fun working on it!

My Collaborative Web Chrome extension was kind of a project that came out of no where. I was feeling kind of bored with work, and wanted something new! This project was a lot of fun to work on, as it was great to revisit some of the basics of web dev and learn more about Go! Excited to use this new knowledge in future projects!

Zoup is a python library inspired by the language Sage's group capabilities, utilizing groups as a main data type. This is still a work in progress, but down the line I would like to add a lot more capabilities so that it is comparable with Sage, and hopefully eventually get to graphic generation, similar to JuliaPoo's Cayley Graphs and pretty things. I would also like to start getting it running like a full-scale operation with more utilization of github actions, version tracking and eventually pip installable. Group Theory is still a new topic for me and I am loving learning as I go!

This project was recently started with my friend Sam and is a work in progress. The main goal of this project is to familiarize myself with the process of data collection, cleaning, and analysis, as well as the adaptation of Neural Networks to real world and more complex applications. I think timeline as of right now is to get the data collection and cleaning done, and then start to work on the model having it prepared for next season, using next season as validation, and then betting the season after if we are satisfied with the results. I do think that as we go we will learn to, and want to automate the features including data collection and training, turning the project into a full scale CI/CD pipeline. It would be cool to have this as a widget or something, where it can optimize bets at the click of a button lol.

Since learning about Linear Algebra, I've been fascinated by embeddings and dimension reduction as a means to understand the data we are handling. Jordan sent me the TensorFlow Word Embedding a while back and it has been such a source for inspiration. As I get more interested in data science, I think understanding what PCA and t-SNE are beyond just what they mean is a huge breakthrough for me! Overall this is jus a fun little notebook I was playing around in, and learned a lot with. Stoked to have done it and learned from it!