The Piste Guide to R and Snowflake

Working with skiLift and skiPatrol

End-to-end guide for R users — Snowflake platform basics, desktop IDEs, Workspace Notebooks, DBI connectivity, and the skiPatrol ML platform.
Author

Simon Field

Published

May 29, 2026

Keywords

snowflake, R, RStudio, Posit, VS Code, workspace notebooks, skiPatrol, skiLift, mlops

Welcome

This guide helps R users run statistical modeling, forecasting, and ML workflows on Snowflake — from RStudio, Posit, VS Code, or Workspace Notebooks — using three open-source packages.

Where to Start

If you are new to Snowflake, start with the Platform Primer and Glossary. If you are new to R on Snowflake but know Snowflake already, jump to Architecture or Local R Setup.

Some chapters are still to come while the guide is reworked following the packages’ rename; links to them are marked with a dagger (†) and are not active yet. See Introduction for what is covered so far.

What You’ll Learn

Key Topics Covered
  • Snowflake primer — Warehouses, containers, ML services (for R users new to the platform)
  • R from your IDE — RStudio, Posit Native App, VS Code; skiLift vs Posit odbc + Snowflake ODBC driver
  • Workspace Notebooks — UX overview, bootstrap, EAI, %%R, CRE, ML Jobs
  • skiLift — SQL REST API, DBI, dbplyr, Arrow (optional ADBC in CRE for fast bulk I/O)
  • skiPatrol — MLOps, Feature Store, Model Registry, Tasks, SPCS, monitoring

The Three Packages

Package Role Repository
snowflake-notebook-multilang Bootstrap R in Workspace Snowflake-Labs/snowflake-notebook-multilang
skiLift DBI via SQL API posit-dev/skiLift
skiPatrol ML platform from R posit-dev/skiPatrol

How to Use This Guide

Your Goal Start Here
New to Snowflake Platform Primer
RStudio / Posit / VS Code Local R Setup → skiLift vs ODBC
Workspace Notebooks Workspaces overview†
Understand SQL API skiLift SQL API
MLOps on Snowflake MLOps chapter
Feature Store deep dive Feature Store Implementation Guide
Term lookup Glossary

Prerequisites

  • Snowflake account (Workspace optional depending on path)
  • R 4.2+; Python 3.10+ for skiPatrol locally
  • See Introduction for disclaimers
CRE speeds Workspace startup (same idea as R)

A Custom Runtime Environment† pre-bakes micromamba, R, %%R, skiPatrol, skiLift, and optionally ADBC (adbcsnowflake) into the notebook image — so analysts skip the cold setup_notebook() install and Go-compiler bootstrap for Arrow I/O. Platform teams own the image; analysts attach cre@<org_name> in notebook settings. Same reproducibility principles as pinning package versions in a CRE profile.

Revision History

Version Date Summary
0.1 2026-05-29 Initial scaffold
0.2 2026-05-29 IDE chapters; chapter renumber
0.3 2026-05-29 Platform primer, glossary, MLOps/SQL API chapters; feedback pass