Completed
[] Tests: Sanity check for the model
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Declarative MLOps: Streamlining Model Serving on Kubernetes
Automatically move to the next video in the Classroom when playback concludes
- 1 [] Musical introduction to Rahul Parundekar
- 2 [] LLMs in Production Conference announcement
- 3 [] Purchase our Swag shirt!
- 4 [] Declarative Paradigm
- 5 [] Why now?
- 6 [] It's great for scalability
- 7 [] Most MLOps tools work well with K8s
- 8 [] Easy-deploys with tool-provided CRDs
- 9 [] Caveats
- 10 [] This talk
- 11 [] 3 Ways to Serve ML Models
- 12 [] Way 1: Serving a Model with an HTTP Endpoint
- 13 [] Way 2: Serving the Model with a Message Queue
- 14 [] Way 3: Long-running Task that Performs Batch Processing
- 15 [] Buil your own container
- 16 [] The main predictor 1/2: Singleton with load method
- 17 [] The main predictor 2/2: Predict
- 18 [] Way 1 5 steps
- 19 [] Way 2 2 steps
- 20 [] Way 3 2 steps
- 21 [] Tests: Sanity check for the model
- 22 [] Bringing it together: Entrypoint
- 23 [] Continuous Integration CI
- 24 [] Create docker-compose.yaml to make it easier for CI
- 25 [] On PR: Run tests with Github Actions
- 26 [] Branch-protection
- 27 [] On PR: Github Actions automatically runs our test
- 28 [] On PR: PRs can be then merged on approval
- 29 [] Container Repository
- 30 [] Continuous Integration CI
- 31 [] On merge to main
- 32 [] Actions that can constraint
- 33 [] TODO
- 34 [] Continuous Delivery
- 35 [] Argo CD
- 36 [] Image promotion with Kustomize
- 37 []
- 38 []
- 39 []
- 40 []
- 41 []
- 42 []
- 43 []
- 44 []
- 45 []
- 46 []
