Thomas Little

Project Portfolio

About

Hello! This is a portfolio of my projects from the past few years, predominantly focused on machine learning (ML). Within this collection, you will find a range of engaging ML projects, each delving into a unique branch of the field. Through these projects, I have documented my immersive learning experiences while applying ML concepts to solve intriguing problems.

Projects

Click on each project to learn more about the problem, the approach, and the results.

• Using Graph Neural Networks for Polymer Prediction Tasks

• This study explores an ML approach to the analysis of block copolymers by harnessing graph neural networks. By replacing the traditional computationally intensive self-consistent field theory simulations with an ML model, we investigate the tradeoff between faster computation times and accuracy.

• Using Transformer Intervention to Increase Compositionality in Small Language Models

• Over the last several years, transformer language models have continued to boast massive training corpuses that are increasingly out of reach for individuals and smaller research teams. To address this issue, this project explores the use of transformer intervention to increase the compositionality of smaller language models, in hopes of improving their performance on downstream tasks and reducing the amount of training data needed for adequate performance.

• Ground Motion Modeling using Deep Neural Networks

• In the field of earthquake engineering, traditional ground motion modeling methods are often limited by the input variables defined by the creator of the model. By harnessing the power of machine learning, this project creates models with customizable inputs that can outperform traditional methods, enabling more adaptable hazard predictions.

• Nonlinear Structural Analysis of Rocking CLT Shear Walls

• Due to emerging sustainability initiatives, the construction industry is steadily expanding beyond traditional "high-strength" structural materials such as concrete and steel. One growing material is cross laminated timber (CLT), which has many structural applications. The area of interest for this study is CLT's potential in high seismic activity through the study of post tensioned (PT) rocking walls. To facilitate this research, computational models were constructed to analyze the nonlinear behavior of the wall under cyclic loading.

Certifications & Education

• Google Cloud Certification - Professional Machine Learning Engineer

• Certification ID: Online Verification
  Issue Date: November 11, 2023

• Stanford University

• MS, Structural Engineering - 2023
  Board Member - EERI Student Chapter

• California Polytechnic State University, SLO

• BS, Architectural Engineering, Magna Cum Laude - 2021
  Minor, Construction Management - 2021
  Board Member - SEAOC Student Chapter
  Academic Tutor