Designed and launched various components within Indeed’s messaging platform. Project consisted of moving away from a monolithic system to building a new microservice leveraging MongoDB to persist conversation read cursors and unread message counts for all users on the platform.Led development and launch for new services handling interview scheduling automation. Projects included work to detect and disable inactive employers to reduce the rate of "employer no-shows" for automatically scheduled interviews by >23% and enabling SMS reminders for various stages of an interview’s lifecycle.

Designed and implemented A/B testing framework within existing supply chain systems while working across multiple teams. Piloted experiments using the new framework impacting multiple core features of Amazon’s supply chain providing savings totaling more than $450mil YoY within its first year of launch.Led team as part of SVP-goal to deprecate multiple legacy web tools used for tracking inventory and replaced them with a single application leveraging ReactJS and serverless AWS offerings. Migrated approximately 35,000 internal users to the new application.Mentored junior engineers working on building ETL job automation framework for ongoing experiments. Time to analyze experiments reduced from 4-6 weeks to 1-2 weeks.

Worked on the Fulfillment Availability team responsible for receiving and aggregating supply and demand signals from Amazon.com and its warehouse network to maintain up-to-date inventory data in order to take and plan customer orders. Designed and implemented highly-available distributed systems running on AWS.Attended Amazon's Machine Learning University (MLU) starting in the summer of 2018.

Migrated all of eLynx's historical data collection (hundreds of terabytes) from on-premises SQL storage to Azure Table Storage (NoSQL). The process involved creating scripts to distribute the data for migration across multiple Azure VMs while having limited access to the production databases due to existing customer traffic and the requirement not to significantly impact server load. Began rewriting data science team's code to follow modern software design patterns and practices to allow the code base to remain maintainable.

Working under supervision of Dr. Roger Mailler, developing an automated system in Java to track objects orbiting in the Earth's atmosphere. Project had large existing code base written entirely in Java. The system collects weather data on remote systems and sends the data to a central hub for processing.

Designed, constructed, and configured a miniature cluster of Raspberry Pi computers as a tool for live demonstrations to illustrate the principles behind how supercomputers work and their capabilities. The cluster runs a chess engine which individuals can interact with via a touchscreen peripheral. The cluster had LED arrays on PCBs which were connected to the GPIO pins of each Raspberry Pi in the cluster. Every node of the cluster used an optimized custom-compiled Gentoo Linux kernel. The cluster also used Slurm resource manager for its scheduler. Assisted in the development of a framework to simulate physical circuit board networks. The simulation used MPI in order to simulate the electrical signals sent and received by the various boards in the network. The purpose of this was to assist a research project at the University of Oklahoma which was working with autonomous networks and highly connected networks. After December of 2015, my role changed to be more of a fill-in for a system administrator until a new System Administrator was hired. Responsibilities during this time included configuring, testing, and pricing provisioning software for Tandy Supercomputing Center. Products included Bitium, Okta, OneLogin, and Azure AD. Also responsible for configuring SSO for the applications used by the employees of TSC. Gained experience with SAML authentication, ADFS, AD, LDAP, and Microsoft Windows Server 2012.Provided user training on how to utilize the Tandy Supercomputer at University of Tulsa at various research facilities, ranging from getting started with C programming and MPI to using tools created by TSC for various software available on the cluster. Designed and implemented new inventory tracking system using Atlassian Jira and custom QR barcodes.

Human Genome Repeat Clustering: Analyzed repeated data sequences found throughout the human genome using the tool call RepeatMasker. The focus of the research was on clustering the repeats into possible groupings based on an accuracy parameter, detecting the repeats which have been defined and characterized by biologists and applying the appropriate labels, and finally producing groups for the remaining repeats which remain to be analyzed by biologists. The project utilized HPC to process the large data sets with Perl and C++ being the languages used on the project.

Developer responsible for the testing suite used at Waterfield Energy to test front-end web UI elements and functionalities. The project was written in Python relying heavily on Selenium WebDriver to automate a chrome window and replicate the testing procedures previously carried out by QA employees. The project required an error reporting system as well as automatic installation for new users and future updates on Windows and OSX platforms. Updates were rolled out using a batch script pulling newest commits to the Git repo the software was staged on. The software maintained the ability to recover from unexpected behaviors of the Waterfield application and reporting to the QA department any bugs which needed to be written up.During development, user training was also carried out to demonstrate new features and functionalities of the software to the rest of the QA staff and how they should use them.

Developed a real-time display of the Tandy Supercomputer’s resource usage and network traffic using Python and the MatPlotLib package to create the graphs. The data was handled by polling a Zabbix server and with the finished product being displayed on a digital signage webpage using AJAX calls to update the image.Ported OU’s “Supercomputing in Plain English” exercises to run on the Tandy Supercomputer. The documentation was likewise converted to describe to new users how to get started using the cluster on the external TSC Wiki. During this process, I developed a Monte Carlo exercise using MPI to test the functionality of the newly constructed cluster. Configured a Hadoop cluster using virtual machines which ran on a VMWare vSphere production server. The purpose of the cluster was to test the efficiency of MapReduce and its possible applications on the Tandy Supercomputer.Configured a DrQueue cluster to test the capabilities of the Tandy Supercomputer to act as a production-level render farm. This included taking created projects from Adobe’s After Effects, Maya, and Blender and rendering each frame of a sequence using a batch scheduler.