Navigating the MLOps Landscape

[ljvmiranda921]

MLOps landscape is confusing: so many tools popping up everyday This blogpost won't present a solution, rather just a field-guide on what to consider

• Talk about the players: cloud providers, end-to-end ML, customized tooling
• Adapt your data science process to the tool? or Adopt tools based on your data science process?
• Build or buy? the case for building in-house ML tools (lessons learned)
  • If it's not your main line of business, don't bother (unless you'll spin it off into its own startup).
  • Exceptions to the rule: you have a dedicated MLE team that will do just that (go-jek feast, Mckinsey QB Kedro)
• Last but not the list, standard due diligence applies. If it's open-source, check the commit activity. Check if there's a roadmap

[ljvmiranda921]

Get info from the following channels:

• MLOps community
• Scale conference
• Talk to some MLEs