LLMs in R

Can R tame the LLM chaos? Let's find out. 🤖

Two talks. Two live demos. One evening that might change how you think about LLMs in R.

Refreshments and networking from 18h00.

Agenda
~~~~~~~~~~~~
* 16h30 – 16h45 | Welcome and News - Handre Williams & Luis de Sousa
* 16h45 – 17h20 | Handre Williams: R as the orchestration + statistical validation layer for LLMs
LLMs are powerful but unreliable. R is rigorous but underestimated. This talk explores what happens when you combine the two — using R as the orchestration backbone and statistical validation layer for LLM workflows. Handre will show practical patterns for calling, chaining, and stress-testing LLMs from R, and how classical statistical tools can keep your models honest in production. Includes a live demo.
* 17h20 – 17h55 | Luis de Sousa: RAG (Retrieval-Augmented Generation) in R
LLMs hallucinate. The fix? Give them better memory. This talk dives into Retrieval-Augmented Generation (RAG) in R — combining DuckDB's vector search capabilities with R's data ecosystem to ground LLMs in domain-specific knowledge. Luis will show how to build a RAG pipeline from scratch, turning your own data into a knowledge base that makes LLMs actually useful in practice. Includes a live demo.
* 17h55 – 18h00 | Wrap up, questions & close
* 18h00 | Refreshments & networking — continue the conversation in person

Details
~~~~~~~~~~~~
When: 14 April 2026 @ 16h30 to 18h00 (GMT+2)
Where: Microsoft Campus in Bryanston, South Africa
Location: https://goo.gl/maps/4ffcC3N6T7S2
Keen to speak at a future event? Submit your session abstracts at https://sessionize.com/johannesburg-r-user-group/

Stay connected
~~~~~~~~~~~~
* Website: https://www.rusergroup.co.za
* Meetup: https://www.meetup.com/joburg-r-users-group
* Twitter: https://twitter.com/JHBRUserGroup
* GitHub: https://github.com/rusergroupcoza
* WhatsApp: https://bit.ly/3CeDifS