MLOps: Value and Limits

Building Effective and Thriving Machine Learning Teams - David Tan & Dave Colls

Topic 8 of 9

Published Jan 27, 2025 · Clip 49:14–52:46 (3:32)

David Tan frames MLOps as another tool that abstracts complexity: an ML platform let data scientists write plain Python without knowing AWS or Kubernetes, and experiment tracking ran on every pull request to compare against champion models. He warns of tool sprawl and vendor coupling, pointing to ThoughtWorks' guide to evaluating ML platforms. David Colls notes MLOps automation needs proper testing to move fast safely, and David Tan adds the key book message: 'you can't MLOps your problems away'—MLOps won't write tests, talk to users, or keep code factored.