ABOUT ME

I am a computer scientist with a concentration in Software Engineering; I also have a background in IT and neuroscience. I am currently working toward becoming a Machine Learning Engineer, and hope to use my skills to benefit humankind.

Check out the About page for more details.

MY PROJECTS

• MLDash

  
  

  I developed the MLDash application for my Computer Science Capstone project. MLDash is a Jupyter Dash app that allows a user to tune machine learning hyperparameters for a reinforcement learning algorithm. Hyperparameters and performance metrics are stored in a MongoDB database, then displayed in the dashboard as a data table and data visualization graphics.

  
  

  The MLDash project was written in the Spring of 2023. As an aspiring machine learning engineer, and software engineer by discipline, I recognize the value in making complicated tasks more accessible to the common user. Training machine learning algorithms requires tuning their (sometimes numerous) hyperparameters and hoping that the changes made boost model performance, while also keeping track of the changes that have been made. The main purpose of this project is to simplify this task for users, automate record keeping, and generate data visualizations that can assist users in the selection of more optimal hyperparameters. While the project currently hosts a single AI algorithm, OpenAI Gym's Cartpole-v1, I anticipate that additional models can be added as modules over time. The algorithm used to solve the cartpole problem is a deep-Q reinforcement learning algorithm that solves the problem of balancing a pole attached to a cart that moves on a frictionless rail. The model acquires data from the environment (observation space) and determines which sequence of actions results in better performance for each situation. The model trains itself by replaying experiences it acquires, prioritizing samples of its experience based on how different those experiences were from its predictions. The model begins training with a higher likelihood to explore, allowing it to gather new experiences by trying different action sequences; as training continues, the model relies less on exploration and more on the exploitation of the experiences it has gathered.

  Navigate to my Portfolio page to learn more.

• More coming soon...

  
  
  

  DataFlow

  DataFlow is an ongoing project that seeks to automate many data analysis tasks under the umbrella of a single dashboard application. The most aggressive ambition for the DataFlow project is to use an AI algorithm to assess imported data, criteria and goals to identify the most relevant modeling approaches, displaying their known strengths and weaknesses to the user. As of December 2023, I have obtained the Google Advanced Data Analytics Professional certificate. This data science track was an excellent experience which helped me to refine my skills in data analysis, data visualization, and business communication. Some of the work completed is worth showing here in my portfolio site but my newfound skills will also inform the development of MLDash 2.0 (Additional OpenAI Gym modules, more advanced data visualizations, greater degree of modularization, improved security, and a more polished user interface), with plans already in the works for MLDash 3.0 (dockerizing the app, to allow for platform independent use and cloud-based parallel processing).

CONTACT ME