/ Home / NuvoMl

NuvoML vs MLFlow

MLFlow:

• open source python library
• can track any number of models
• Can easily create api link for the models
• Deployment and model version tracking
• app/UI needs to be created by user. Just the API link will be provided by MLFlow
• Server cost/maintanance should be taken care by the user
• No user anayltics

NuvoML:

• Free and subscription based service
• Can track limited number of models based on subscription
• No seperate API link
• No Version tracking
• app/UI is created by NuvoML and various themes can be used for deployment
• Server cost/maintainance is taken care by featurepreneur
• No user analytics(as of right now)

Future plans/ ideas:

• Automatic theme converter - get the html/css/javascript from frontend developer and convert it to our required format
• User analytics(prediction parameters given by end-user, ip of end-users)
• seperate API link for postman/curl

• deployment version tracking and changing version on the fly (v1 to v2 with just the click of a button)

• (Maybe) python library for local model tracking(can track any number of models in the local) with UI similar to MLFlow and integration with NuvoML (deploy/change version, etc. on the local UI itself)

  • featureprenuer login creds should be specified to deploy, otherwise can be just used for model tracking.
  • Can track several models in the local. When they need to deploy it, just a click in local will deploy in NuvoML and give the deployment link
  • Can switch the current production model in the local UI itself.