Estimating the cost of an ML project

Hi there, I’m planning to build an audio recommendation system that we want to deploy in production. How can I estimate how much compue cost will I incur, say on a monthly basis? (Including ongoing model monitoring, retraining, and re-deployment)

It depends on what infrastructure you use, you can check cloud services for the compute power, memory and other services you are planning to use how much would that come up to. Try AWS, IMB, AZURE, GOOGLE CLOUD… Building your own infrastructure is a rather expensive choice I think.

Thanks, I see. What would be an alternative path to building our own infrastructure? (if this was to become a commercial app eventually, though right now it’s just a prototype)

Maybe, depends on scale of business
and its prospect, bur building you own infrastructure could be quite expensive!