Research
This page describes my research experience associated with my education. For industry and public sector research experience, please see Work Experience.
Reinforcement Learning Applied to Neural Network Adaptation
November - December 2023
I investigated applying reinforcement learning to neural network adaptation in the context of cluster scheduling. This was done as part of the fall 2023 offering of Reinforcement Learning. This is related to my current research. For more information, please see my report here. The abstract is attached below.
The design of cluster schedulers for deep learning jobs has evolved to take into account neural network training behavior. We believe there is an opportunity for schedulers to adapt model architectures based on resource availability. In this work, we present a reinforcement learning agent that informs a neural network trainer on how to increase the complexity of a DNN during training so as to produce a better final model. This would be used in scenarios where a deep learning job receives extra GPUs during training. Our initial evaluation shows that our agent is able to provide good adaptation policies at different stages of training but not for different starting models.
Alibaba Trace Analysis
January - May 2023
Linda Zhao and I conducted a visual analysis of the Alibaba trace dataset. This was done as part of the spring 2023 offering of Debugging Cloud Computing under the supervision of Professor Raja Sambasivan and PhD student Darby Huye. For more information, please see our report here. The abstract is attached below.
Microservice design of cloud applications has been an area of academic research for some time. This has been motivated by more and more cloud applications being implemented as collections of lightweight microservices run on top of distributed systems. In this work, we seek to obtain a better understanding of Alibaba’s microservice architecture via an analysis of the well-known Alibaba trace dataset. Our work provides some preliminary analysis of the architecture as well as identification of the various problems in the dataset that would need to be addressed in future research on this topic.
Fluid Solver for Incompressible Flow in a Rectangular Domain with Arbitrarily-placed Inclusions
August 2021 - May 2022
For my senior honors thesis, I worked with Professor Thomas Fai (Brandeis Department of Mathematics) to develop an algorithm to approximate fluid velocity in varying spatial domains using the Navier-Stokes equations. The abstract of my thesis is attached below. My thesis is available here. I also have posted a template for future students to help them write their thesis in a format following department requirements here.
Microfluidic devices are used to process fluids at a small scale. Some of these devices contain physical
obstacles that are used to manipulate the flow of fluid through the device. In our work, we implement a Python program for approximating the velocity of an incompressible fluid over a rectangular domain using the Navier-Stokes equations. Our strategy, which is built primarily on common finite difference methods, allows users to easily add obstacles into the domain and observe their effects on fluid flow. In the first half of the thesis, we describe the derivation, implementation, and benchmarking of our strategy. In the latter half, we describe some numerical experiments that we performed with our program to observe the effects of inclusions on both fluid flow and hydraulic resistance. We compare the hydraulic resistance reported by our program with an analytic approximation. Lastly, we briefly mention some possible expansions on our work.
AAU Higher Education Job Postings Visualization Tool
January - May 2020
For my 2020 spring semester independent study, I worked with two undergraduate students to develop a web application that allows users to study higher-education job postings in a variety of modes. The project was advised by Prof. Antonella Di Lillo and computer science PhD student Solomon Garber. Our team’s work was done in collaboration with Dr. Jessica Liebowitz as part of the AAU PhD Education Initiative. Our goal was to create a public online tool for people to use in their search for jobs and careers. Our web application allows users to view job counts and top skills based on career area, job title, job type, and additional fields. The users make a variety of input selections and the server returns the results to be visualized client side from a dynamically constructed SQL query.
