Rainbow Six Siege Analysis
With the idea of helping players find a better playstyle, I started developing this project with the hope of training a ML model on player behavior to suggest better positions or find a better playstyle. I currently developed how I want to organize the player data from the R6Stats API. As for future plans, I'm thinking about how to develop the ML model and store the data from profiles for the model to use. You can find the repository here: R6 Analysis Repository
Classification Using Graph Convulution
This project was a replication piece in bringing a strong production model of Graph Convolution to Pytorch. I focused on building my own convolution layer so that I could classify nodes in a graph dataset. I also conducted hyperparameter testing to see how each hyperparameter affects the accuracy and speed of the simple model that replicates the original work of Kipf/Welling. This gave me a good baseline in studying the advantages and disadvantages of Pytorch and Tensorflow when running the same model. The Github Repository can be found here: PYGCN Repository.
TrustWorthy Modules
This project was focused on creating an implementation on scoring NPM modules and deploying that to the cloud for a company to see what modules may be deemed “trustworthy”. This project involved having to create the initial grading scripts based on aspects within the NPM modules and then building a Flask API around that module. Once this was done, we had built our own CI/CD pipeline to create a DevOps cycle, where any change was tested, reviewed, and deployed to Google Cloud after approval by the other team members. Unfortunately, I can't display this project because it's tied to a Github Classroom repository. If you would like to see this project, shoot me an email!