Background: The company was founded in 2015 and had a legacy Node.js stack running version 0.10 of Node, with frontend being delivered using Jade templates. In order to more quickly iterate on web development projects, I proposed and executed the project to migrate the application to modern Node.js and a React.js frontend.

Background: Deploying an HPC cluster in the cloud was originally done by utilizing the vendor-specific solutions for each cloud, i.e. AWS ParallelCluster, Azure CycleCloud, and Google Deployment Manager. In order to quickly adopt new features and iterate faster with a small team, we redesigned Parallel Works to create a uniform cluster provisioning process utilizing Terraform on all three major clouds. This feature is now the centerpiece of the Parallel Works product, currently in use by the government organization NOAA for their R&D.

Background: Most of the resources created within our AWS environment were architected from the console. This makes it hard to repeat the process when other business units want a similar solution and hard for new team members to understand the scope of a solution. In order to solve this, we want to introduce CloudFormation and CDK options to the business units using the cloud. However, it can be quite a bit of leg work to create a resource that is compliant with the various security policies required.

Background: During my time in this position, the HPC had over 5,000 compute nodes. With such a large amount of computers, and a small team, it can be hard to keep track of everything. In order to remediate this, I created a software which aggregated data from all 5,000+ machines into a single Node.js cache server and backing database.

Background: While the data in the information aggregator I created was easier to access than previously, it still required users to access it via API, SQL, or GraphQL. It would be easier if there was some visual way to interpret the information. Because of this, I created a React.js application which was a visual representation of our data center, as if you were looking at it from a top-down view. You can drill down into individual racks, and then into servers, and see real time information about those servers. The views of the servers are color coded by either power usage, CPU usage, or temperature, so you can easily identify racks or nodes with outliers.

Background: Their marketing website was hosted on WordPress and had been hacked repeatedly, creating a negative image for the company. They asked if I had any ideas and I proposed switching the site to use Gatsby.js, as statically generated sites are inherently more secure.

Background: As part of their business, Foundation Software mails paychecks to thousands of customers per week. The process of printing the labels for this was done with Microsoft Word and folders. This project replaced that process with an automated system.