David Earnest

Hi! I'm David.

I'm interested in building exciting software products that improve people's lives. I have worked for Siemens PLM Software and Northrop Grumman.

Email  /  CV  /  Github

Projects

I have completed a few projects on topics that interest me. I'm broadly interested in machine learning, optimization, and software development.

A webapp that hosts an iGPT model that I trained on MNIST. iGPT is an autoregressive model that uses masked self-attention to predict the next pixel in an image given previous pixels. Once that model is trained, we can sample new images one pixel at a time, similar to how LLMs work for text. The backend is in Python, and the frontend is in HTML, CSS, and JS.

An interactive program that lets you draw digits (0-9) with your mouse and send them to a trained LeNet model, which predicts the digit given the raw pixels.

A simple checkers AI that will play against you using the Minimax algorithm. The approach here is roughly similar to IBM's Deep Blue chess AI.

A simple project that I did for a class in evolutionary algorithms. I tested the benefits of treating the population of an EA as an ensemble for classification tasks.

A simple class project that I did for my deep learning class. A bidirectional LSTM and U-Net style model to predict proton-proton collision points in LHC experiments.