Getting Started with LLM APIs in R

Building AI-powered applications with {ellmer}

AI
LLM
Beginner
Author

Sara Altman (Posit PBC)

Overview

Beginner Friendly AI/LLM Shiny

LLMs are transforming how we write code, build tools, and analyze data, but getting started with directly working with LLM APIs can feel daunting. This workshop introduces participants to programming with LLM APIs in R using {ellmer}, an open-source package that makes it easy to work with LLMs from R.

What You’ll Learn

  • 📡 Calling LLMs from R - Basic API integration and response handling
  • 🎯 System Prompt Design - Crafting effective prompts for specific tasks
  • 🔧 Tool Calling - Enabling LLMs to execute R functions
  • 💬 Building Chatbots - Creating interactive conversational interfaces

Prerequisites

Required Knowledge:

  • Basic R familiarity
  • No AI or machine learning background needed

Setup:

  • R environment with internet access
  • API keys (will be provided during workshop)

Key Packages & Tools

{ellmer}

{shinychat}

OpenAI API

Anthropic API

Workshop Content

1. Introduction to LLM APIs

Understanding how to interact with large language models programmatically:

  • API authentication and configuration
  • Request/response structure
  • Token management and costs
  • Error handling and best practices

2. The {ellmer} Package

{ellmer} provides a unified interface for working with multiple LLM providers:

library(ellmer)

# Connect to an LLM
chat <- chat_openai(
  model = "gpt-4",
  system_prompt = "You are a helpful R programming assistant."
)

# Send a message
response <- chat$chat("How do I read a CSV file in R?")

3. System Prompt Engineering

Learn to design effective system prompts that guide LLM behavior:

  • Defining role and expertise
  • Setting tone and style
  • Providing context and constraints
  • Examples of good vs. bad prompts

4. Tool Calling

Enable LLMs to execute R functions and interact with your data:

  • Defining tool schemas
  • Registering R functions as tools
  • Handling tool execution
  • Multi-turn conversations with tools

Example use case: LLM that can read files, perform calculations, and generate plots.

5. Building Basic Chatbots

Create interactive conversational applications:

  • Using {shinychat} for UI
  • Managing conversation state
  • Streaming responses
  • Adding context and memory

Practical Applications in Pharma

  • 📊 Data exploration assistants - Natural language queries on clinical data
  • 📝 Report generation - Automated narrative generation from analysis results
  • 🔍 Code review helpers - Explain complex statistical code
  • 📚 Documentation assistants - Generate function documentation

Workshop Materials

NoteResources

Workshop Link: https://skaltman.github.io/r-pharma-llm/

GitHub: https://github.com/posit-dev/ellmer

Instructor: Sara Altman is a Data Science Educator at Posit PBC, focusing on making AI tools accessible to R users.

Example: Simple LLM-Powered Data Assistant

library(ellmer)
library(dplyr)

# Create a chat interface with tools
chat <- chat_openai(
  model = "gpt-4",
  system_prompt = "You are a data analysis assistant. 
                   Use the provided tools to answer questions about data."
)

# Register tools
chat <- chat %>%
  register_tool(
    "summarize_data",
    function(data) {
      summary(data)
    },
    description = "Get summary statistics of a dataset"
  )

# Use it
response <- chat$chat(
  "Can you summarize the mtcars dataset?",
  data = mtcars
)

Learning Outcomes

By the end of this workshop, you will be able to:

✅ Set up and configure LLM API connections in R
✅ Design effective system prompts for specific tasks
✅ Implement tool calling to extend LLM capabilities
✅ Build basic chatbot interfaces with {shinychat}
✅ Understand best practices for LLM integration in pharma workflows

Next Steps

After this workshop, consider:

Additional Resources

Similar Workshops

Tools & Resources

Next Steps

Last updated: November 2025 | R/Pharma 2025 Conference