

Timeframe: 02/2021 – 06/2021
Position: Fullstack Engineer (contract)
Technologies: JavaScript, Phaser, HTML5, CSS, NodeJS, Web Sockets, AWS Lambda, AWS Route53, AWS S3, AWS Transcribe, AWS EC2, AWS Certificate Manager
Status: Released internally to SNHU
The Story:
My last project with Southern New Hampshire University. It was also pretty much the culmination of all of the research that had been done up to that point with the other game projects.
The web-based app was a 3-way push-to-talk style audio chat app, with a visualization component that allowed users to see both the relative volume of those talking (illustrated on the right, though the screenshot is of the Libra Text app, since the audio version screenshots are lost to the sands of time- and bad hard drives), and how much of the conversation was taken up by users on the left. This was implemented to help users prepare for interviews remotely, because this was still the pandemic times.
The audio was processed as push-to-talk in order to conserve on computing resources that would otherwise be needed to process real-time audio. We sent WAV files in chunks to a NodeJS server, which then transmitted them to the other users, and also sent a copy to AWS Transcribe, which would create text transcriptions of the audio and store them in an S3 bucket. That, in turn, would then be processed by a Machine Learning model (provided by another team) to provide the user with insights into their speech during mock interviews.
Phaser was used mainly to allow for quick and easy resizing of the web app in the browser window, because the window would be resized if there were other apps on the screen that were used during sessions, and Phaser handles this far better than HTML resizing does!
I serviced the project end-to-end, spec’ing out the AWS services, and then implementing both front-end web app and back-end NodeJS server. A secondary team of a Machine Learning expert and a full web-development team then handled the output transcriptions for a separate app.