Inspiration
We saw that the time, effort, and money required to use machine learning could easily overwhelm those who are less capable. Therefore, we wanted to build something with ease of use and low cost in mind.
What it does
There are two modes of operation on our platform. For the first, or local mode of operation, the user brings their own machines and connects them to the platform. We envision this use case for universities that have have excess computing power that would otherwise be spend idling. The second mode is where you use our GPU accelerated server for your training. We envision this use case for companies that lack the resources to setup their own GPU servers.
How I built it
The main platform is built with pure PHP for the back end and Bootstrap for the front end. The local desktop client is written in C++ .
Challenges I ran into
The whole LAMP stack was giving us problems that took us a long time to fix.
Accomplishments that I'm proud of
A very efficient method of dispatching jobs to clients.
What I learned
MySQL has a bug with how it handles sudoers vs non-sudoers
What's next for MassMl
Refactor the distribution code for both server and client. Get repository on GNU Savannah.
Log in or sign up for Devpost to join the conversation.