I didn't like the course registration process and thought we could do better.

We're exploring the limits, intricacies and implications of Clusterpoint's document-oriented database as a service and solving a very practical problem along the way.

Share this project:


posted an update

As a college student, I can tell you completely unequivocally that course registration sucks right now. Procedures can be absolutely aggravating for students, get drawn out over periods of over a month, and in the end, the solution is tragically suboptimal. Professors pour endless hours into improving the educational experience for their students, but none of that matters if we can’t register for the right classes.

At it’s root, course registration is just one example of what game theorists call a marriage problem: Back in 1962 Lloyd Shapely won a Nobel prize with his solution to these problems, but at the time, applications of his work seemed largely impractical. Today, thanks to Clusterpoint’s API and technological innovation, his work can become a reality. By next year, or next registration cycle, or even today, here’s what the course registration experience could look like:

Students are free to explore classes through whatever medium their colleges already offer, whether than means course catalogues, talking to friends or auditing classes. Then, they can come here, scroll down, input preferences, and that’s it.

If preferences change, all a student has to do is resubmit, and the entire system gets re-optimized. We’re still building the heuristic around the values that colleges currently hold: senior priorities, making sure everyone can fulfill their requirements, etc but as classes fill up and students get put on waitlists, one student’s decision necessarily affects everyone else.

What I really want to highlight, is that rather than building an app and using Clusterpoint on top as a generic database, we really wanted to start by understanding the unique features Clusterpoint offers, using that as a foundation for our design.

Document-oriented databasing means built-in full-text search in milliseconds with no integration overhead. Database as a Service distribution means that even as students change their preferences, our service can continue providing real time optimizations with minimal latency.

The final point, and here’s one place you might not expect Clusterpoint to shine, Registration happens in a few very sharp peaks throughout the year. It’s not a persistent service but when it happens the traffic can get rough. Unlike other services. Clusterpoint provides instant scalability and a pay-per-use cost model which means even with highly erratic traffic, end users don’t have to worry about under-provisioning or performance penalties.

Our hack isn’t going to save the world, but it is a practical solution to a problem we were facing. Maybe more importantly, it’s providing insights into how the specifications of noSQL alternatives affect so much more than just efficiency. They have a powerful impact on the solutions we’re able to build, and the problems we’re inspired to solve. That’s not as sexy as a chrome plugin, but it is worth fighting for.

Log in or sign up for Devpost to join the conversation.