Meeshkan

Make money by selling your device's processing power at the flick of a switch.

Meeshkan is a revolutionary network that allows you to sell processing time from your device to companies and individuals doing machine learning research.

FOR SELLERS

As Machine Learning becomes more popular, hundreds of thousands of small companies are paying very high bills for computing infrastructure. At the same time, people like you have devices that idle around for hours and could be used to do these computations. Enter Meeshkan! We match people like you with companies. They pay you for your device time, they save money because the market dictates the price, and everyone goes home happy.

So how much will you earn? For an owner of a smartphone or tablet, you can earn up to 70 cents a day. For a PC with an NVIDIA graphics card, this can reach three dollars a day. You can sign up as many devices as you want. Libraries, universities and labs can use this service to make extra income just by turning Meeshkan on.

By participating in this service, in addition to making money, you are helping thousands of individuals small companies around the world undertake Machine Learning projects in an affordable, scalable way.

FOR BUYERS

You've taken the Andrew Ng course, you've set up a few simple Machine Learning problems, you start them off runing and AHHHHH you're depressed and in debt. The jobs are either too expensive, too slow or both. That's no way to iterate! At the same time, individuals have devices that, when pooled together, could run these jobs for you and help you save money. Enter Meeshkan! We match Machine Learning enthusiasts like you with devices that do the learning. You save money because the market dictates the price, device owners make a few bucks for doing more or less nothing, and everyone goes home happy.

So how much will you pay? Meeshkan's realtime bidding system allows you to bid on CPU time anywhere from fractions of a cent to dollars an hour and will match you with machines that fit your bid level. Because your model is saved on each iteration of your model, you can change your bid dynamically depending on your needs. Even at very small bid levels, Meeshkan's parallel processing network allows for reasonable processing time and, most importantly, gives you a tool to explore multiple ML models at once.