OurDataHelps was one of Qntfy's first major products built to quantatively study mental health. It's premise is to descern unique natural language communication patterns representative of mental health conditions by analyzing social media data. Qntfy has conducted its own study to understand, predict, and prevent deaths by suicide since the release of OurDataHelps in 2016.
Each study can be made available for public participation or limited to a set of participants which the practitioner of the study controls. All data is volunteerly provided by participants who can decide how much or little they'd like to provide. Data for each study is collected, processed, and stored securely and when accessed, it is deidentified for protection of the participants and success of the study.
To date, OurDataHelps has been used to study
OurDataHelps owes much of its success to its original creator, James Dennis. Without his efforts it would have been much more difficult to expand the project to as many studies as it did. After Qntfy's study was well underway, James moved on to other projects and I took over his role as project lead and developer.
I was responsible for building, deploying, and managing every custom deployment of OurDataHelps. As the subject matter expert of the product I was the technical pre-sales consultant, built proof of concept integrations and customizations for customers, was the primary point of contact for techincal and other matters for the duration of each study, and ensured all data was delivered securely to each customer regularly.
From a technical perspective, the most challenging aspects of this process were the design and data requirements. While I have a strong interest in frontend, design, and UI/UX, at that time I had not crafted a "pixel-perfect" design to match a customer's specifications and/or their corporate brand before. You can probably see my skills progress across each successive deployment.
With each release came requests for access to data from more social media platforms that we did not support. In total I was responsible for building and integrating five and managing data flows for all eight social media OAuth-based APIs.
As with Quinn, the backend was was designed as a system of microservices. It was optimized to process lots of data quickly, store data securely, and handle deidentification of data easily. It's built with GoLang and Python, supported by Kafka, and deployed on Mesos/Marathon.
It was great to be a part of OurDataHelps. I was able to realize full-stack engineering almost in its entirety and it helped solidify my deep interest in building products from the ground up.