Demystifying MLOps
summary
Claims that Vetiver makes model deployment feel native to data scientists and shows an end to end Vetiver workflow in R and Python. Demonstrates versioning with pins/model board (names, hashes, prototypes), deployment to a local REST API via Plumber (R) or FastAPI (Python) and targets like RStudio Connect or SageMaker, and monitoring with pinned metrics, plotting, model cards and p-type input validation. Shows supported model families (scikit‑learn, torch), custom handler escape hatch, and working live examples.
AI-written summary · about
chapters (4)
related talks
19mDemystifying MLOps
50mBuilding an MLOps strategy from the ground up
27mPractical MLOps for better models
18mDemystifying MLOps with Vetiver
34mHolistic MLOps for better science
26mMLOps for the rest of us- A poor man's guide to putting models in production
13mMLOps at Scale
24mA Hitchhiker's Guide to the MLOps Experience